Skip to content

armgit / eve

Public Imported from GitHub
0

The Open Framework for Building Agents

feat(eve)!: rename AI SDK evaluate to decide (#4426)

Andrew Barba198d6f5cefed
diff --git a/.changeset/ai-sdk-decide-rename.md b/.changeset/ai-sdk-decide-rename.md
new file mode 100644
index 000000000..f97b2acac
--- /dev/null
+++ b/.changeset/ai-sdk-decide-rename.md
@@ -0,0 +1,5 @@
+---
+"eve": minor
+---
+
+Breaking: follows the AI SDK rename from evaluation to decisions and updates to `[email protected]`. `evaluate` from `eve/ai` is now `decide`, and `auto` and `t.judge(...)` accept `Experimental_DecisionModel` instances such as `provider.decisionModel(...)`. Replace `evaluate` imports with `decide` and `evaluationModel(...)` calls with `decisionModel(...)`. Extensions built against the previous `tool` or `dynamicTool` capability must be rebuilt against this release.
diff --git a/apps/docs/lib/geistdocs/redirects.test.ts b/apps/docs/lib/geistdocs/redirects.test.ts
index c08a5def5..791af2d00 100644
--- a/apps/docs/lib/geistdocs/redirects.test.ts
+++ b/apps/docs/lib/geistdocs/redirects.test.ts
@@ -131,6 +131,8 @@ describe("docsRedirects", () => {
["/docs/guides/ucp", "/docs/protocols/ucp"],
["/docs/guides/ucp.mdx", "/docs/protocols/ucp.mdx"],
["/docs/evals", "/docs/evals/overview"],
+ ["/docs/guides/evaluate", "/docs/guides/decide"],
+ ["/docs/guides/evaluate.md", "/docs/guides/decide.md"],
])("redirects %s to %s", (source, destination) => {
expect(docsRedirects).toContainEqual({ source, destination, permanent: true });
});
diff --git a/apps/docs/lib/geistdocs/redirects.ts b/apps/docs/lib/geistdocs/redirects.ts
index 71ad50f15..c65fa8c56 100644
--- a/apps/docs/lib/geistdocs/redirects.ts
+++ b/apps/docs/lib/geistdocs/redirects.ts
@@ -84,6 +84,7 @@ export const docsRedirects: DocsRedirect[] = [
...createDocsRedirects("/getting-started/multiple-root-agents", "/concepts/project-structure"),
...createDocsRedirects("/multi-agent-projects", "/concepts/project-structure"),
...createDocsRedirects("/getting-started/first-agent", "/tutorial/first-agent"),
+ ...createDocsRedirects("/guides/evaluate", "/guides/decide"),
];
export const rootMarkdownRedirects: DocsRedirect[] = [
diff --git a/apps/templates/eve-chat-template/package.json b/apps/templates/eve-chat-template/package.json
index 485519ace..bcfc8f40b 100644
--- a/apps/templates/eve-chat-template/package.json
+++ b/apps/templates/eve-chat-template/package.json
@@ -36,7 +36,7 @@
"@vercel/analytics": "2.0.1",
"@vercel/connect": "^2.4.0",
"@vercel/speed-insights": "2.0.0",
- "ai": "7.0.105",
+ "ai": "7.0.128",
"better-auth": "1.7.1",
"class-variance-authority": "0.7.1",
"clsx": "2.1.1",
diff --git a/apps/templates/eve-chat-template/pnpm-lock.yaml b/apps/templates/eve-chat-template/pnpm-lock.yaml
index b655fcdf2..0b3c9f271 100644
--- a/apps/templates/eve-chat-template/pnpm-lock.yaml
+++ b/apps/templates/eve-chat-template/pnpm-lock.yaml
@@ -43,13 +43,13 @@ importers:
'@vercel/connect':
specifier: ^2.4.0
'@vercel/speed-insights':
specifier: 2.0.0
ai:
- specifier: 7.0.105
- version: 7.0.105([email protected])
+ specifier: 7.0.128
+ version: 7.0.128([email protected])
better-auth:
specifier: 1.7.1
@@ -67,7 +67,7 @@ importers:
version: 0.45.2(@neondatabase/[email protected])(@types/[email protected])(@upstash/[email protected])([email protected])
eve:
specifier: ^0.66.3
lucide-react:
specifier: 1.33.0
version: 1.33.0([email protected])
@@ -126,20 +126,20 @@ importers:
packages:
- '@ai-sdk/[email protected]':
- resolution: {integrity: sha512-c8ztST/CslupqO3p7izsAuG3TlKg9mhE6mGJ42iUADsSMJIV5IOKd9AsLG33QCAY7SWs6NDO2giM/rq7bUTNlQ==}
+ '@ai-sdk/[email protected]':
+ resolution: {integrity: sha512-4lTyt8a4BMfVz4QGo+VO5odakmYMX5kq+xiRgdQGGY7o/jP2WSasp6LvIvYvt/lHzz1MfE+xKTOJhz7mMwjynA==}
engines: {node: '>=22'}
peerDependencies:
zod: ^3.25.76 || ^4.1.8
- '@ai-sdk/[email protected]':
- resolution: {integrity: sha512-gw/bcNseOGSs59TMtV4H1KwqXXe24NHgx+uBYr98pa4Fg6Uvp8hPPxjuckVnfFFyIxHCDew8so0ujpvcSuyMZA==}
+ '@ai-sdk/[email protected]':
+ resolution: {integrity: sha512-amqDxFnw9+dpVrDACRDGmExUUNOH9C1D2bhg/DqWJnamEOieDTXBBXLHujSby/0MqHdAFksnpE8Pzfqx6ABMBw==}
engines: {node: '>=22'}
peerDependencies:
zod: ^3.25.76 || ^4.1.8
- '@ai-sdk/[email protected]':
- resolution: {integrity: sha512-VYMBxIQdcHqbIf1j+YZlI9Ati6LZ4wJe0GGd4z4a5H/KxTggjeOiyaVYTnfF7LHZK5jMQ+rofmzz4QPqf++NUw==}
+ '@ai-sdk/[email protected]':
+ resolution: {integrity: sha512-1Jtmn36VNMOBo9CMUQrPqsR744qtqRloEiXpWCYS8ffPlhS2CrSNmoho5lPjp3R7EnkzBsGI6p51TA7fQJQZdA==}
engines: {node: '>=22'}
@@ -2545,8 +2545,8 @@ packages:
'@workflow/[email protected]':
resolution: {integrity: sha512-pav4F2BoirECWR7Nf1TKt+2eETcBj7jj4cBefQ8VXQCA6NPkaKeLfj/zMgi+3zYV5ZIBT4GuUiphsj0/b9hPQQ==}
- resolution: {integrity: sha512-gkV3W+tTtWwQ41KNV/KjV8wabBrk+bnbnZ3f0AWp/T9LLyljose8uhFViKleVy7tIdEiR2PiS4BpbNdcExf7rQ==}
+ resolution: {integrity: sha512-WMKN83DodYtH52ETPHq0jrSwj3wU0aHkhLHXMrwMRabC/aMpIqeWsXJgjkUwrOqkaKHWONdupk/2ttPRvAyFcg==}
engines: {node: '>=22'}
peerDependencies:
zod: ^3.25.76 || ^4.1.8
@@ -4102,23 +4102,23 @@ packages:
snapshots:
dependencies:
- '@ai-sdk/provider': 4.0.17
- '@ai-sdk/provider-utils': 5.0.43([email protected])
+ '@ai-sdk/provider': 4.0.22
+ '@ai-sdk/provider-utils': 5.0.54([email protected])
'@vercel/oidc': 3.2.0
zod: 4.4.3
dependencies:
- '@ai-sdk/provider': 4.0.17
+ '@ai-sdk/provider': 4.0.22
'@standard-schema/spec': 1.1.0
'@workflow/serde': 4.1.0
eventsource-parser: 3.1.1
undici: 7.29.1
zod: 4.4.3
- '@ai-sdk/[email protected]':
+ '@ai-sdk/[email protected]':
dependencies:
json-schema: 0.4.0
@@ -6045,13 +6045,13 @@ snapshots:
dependencies:
execa: 5.1.1
dependencies:
'@vercel/oidc': 4.0.0
optionalDependencies:
- ai: 7.0.105([email protected])
+ ai: 7.0.128([email protected])
'@vercel/[email protected]': {}
@@ -6068,11 +6068,11 @@ snapshots:
'@workflow/[email protected]': {}
dependencies:
- '@ai-sdk/gateway': 4.0.85([email protected])
- '@ai-sdk/provider': 4.0.17
- '@ai-sdk/provider-utils': 5.0.43([email protected])
+ '@ai-sdk/gateway': 4.0.104([email protected])
+ '@ai-sdk/provider': 4.0.22
+ '@ai-sdk/provider-utils': 5.0.54([email protected])
zod: 4.4.3
@@ -6520,9 +6520,9 @@ snapshots:
dependencies:
- ai: 7.0.105([email protected])
+ ai: 7.0.128([email protected])
nitro: 3.0.260903-beta
undici: 8.9.0
optionalDependencies:
diff --git a/apps/templates/eve-llm-council-template/package.json b/apps/templates/eve-llm-council-template/package.json
index 7d99bf447..0c9518b67 100644
--- a/apps/templates/eve-llm-council-template/package.json
+++ b/apps/templates/eve-llm-council-template/package.json
@@ -18,7 +18,7 @@
},
"dependencies": {
"@comark/react": "0.6.2",
- "ai": "7.0.105",
+ "ai": "7.0.128",
"eve": "^0.66.3",
"geist": "1.7.2",
"next": "16.3.4",
diff --git a/apps/templates/eve-llm-council-template/pnpm-lock.yaml b/apps/templates/eve-llm-council-template/pnpm-lock.yaml
index ea6ca2954..ee0f0208d 100644
--- a/apps/templates/eve-llm-council-template/pnpm-lock.yaml
+++ b/apps/templates/eve-llm-council-template/pnpm-lock.yaml
@@ -12,11 +12,11 @@ importers:
specifier: 0.6.2
version: 0.6.2([email protected])
ai:
- specifier: 7.0.105
- version: 7.0.105([email protected])
+ specifier: 7.0.128
+ version: 7.0.128([email protected])
eve:
specifier: ^0.66.3
geist:
specifier: 1.7.2
@@ -48,20 +48,20 @@ importers:
packages:
- '@ai-sdk/[email protected]':
- resolution: {integrity: sha512-c8ztST/CslupqO3p7izsAuG3TlKg9mhE6mGJ42iUADsSMJIV5IOKd9AsLG33QCAY7SWs6NDO2giM/rq7bUTNlQ==}
+ '@ai-sdk/[email protected]':
+ resolution: {integrity: sha512-4lTyt8a4BMfVz4QGo+VO5odakmYMX5kq+xiRgdQGGY7o/jP2WSasp6LvIvYvt/lHzz1MfE+xKTOJhz7mMwjynA==}
engines: {node: '>=22'}
peerDependencies:
zod: ^3.25.76 || ^4.1.8
- '@ai-sdk/[email protected]':
- resolution: {integrity: sha512-gw/bcNseOGSs59TMtV4H1KwqXXe24NHgx+uBYr98pa4Fg6Uvp8hPPxjuckVnfFFyIxHCDew8so0ujpvcSuyMZA==}
+ '@ai-sdk/[email protected]':
+ resolution: {integrity: sha512-amqDxFnw9+dpVrDACRDGmExUUNOH9C1D2bhg/DqWJnamEOieDTXBBXLHujSby/0MqHdAFksnpE8Pzfqx6ABMBw==}
engines: {node: '>=22'}
peerDependencies:
zod: ^3.25.76 || ^4.1.8
- '@ai-sdk/[email protected]':
- resolution: {integrity: sha512-VYMBxIQdcHqbIf1j+YZlI9Ati6LZ4wJe0GGd4z4a5H/KxTggjeOiyaVYTnfF7LHZK5jMQ+rofmzz4QPqf++NUw==}
+ '@ai-sdk/[email protected]':
+ resolution: {integrity: sha512-1Jtmn36VNMOBo9CMUQrPqsR744qtqRloEiXpWCYS8ffPlhS2CrSNmoho5lPjp3R7EnkzBsGI6p51TA7fQJQZdA==}
engines: {node: '>=22'}
'@comark/[email protected]':
@@ -551,8 +551,8 @@ packages:
'@workflow/[email protected]':
resolution: {integrity: sha512-pav4F2BoirECWR7Nf1TKt+2eETcBj7jj4cBefQ8VXQCA6NPkaKeLfj/zMgi+3zYV5ZIBT4GuUiphsj0/b9hPQQ==}
- resolution: {integrity: sha512-gkV3W+tTtWwQ41KNV/KjV8wabBrk+bnbnZ3f0AWp/T9LLyljose8uhFViKleVy7tIdEiR2PiS4BpbNdcExf7rQ==}
+ resolution: {integrity: sha512-WMKN83DodYtH52ETPHq0jrSwj3wU0aHkhLHXMrwMRabC/aMpIqeWsXJgjkUwrOqkaKHWONdupk/2ttPRvAyFcg==}
engines: {node: '>=22'}
peerDependencies:
zod: ^3.25.76 || ^4.1.8
@@ -855,23 +855,23 @@ packages:
snapshots:
dependencies:
- '@ai-sdk/provider': 4.0.17
- '@ai-sdk/provider-utils': 5.0.43([email protected])
+ '@ai-sdk/provider': 4.0.22
+ '@ai-sdk/provider-utils': 5.0.54([email protected])
'@vercel/oidc': 3.2.0
zod: 4.4.3
dependencies:
- '@ai-sdk/provider': 4.0.17
+ '@ai-sdk/provider': 4.0.22
'@standard-schema/spec': 1.1.0
'@workflow/serde': 4.1.0
eventsource-parser: 3.1.1
undici: 7.29.0
zod: 4.4.3
- '@ai-sdk/[email protected]':
+ '@ai-sdk/[email protected]':
dependencies:
json-schema: 0.4.0
@@ -1155,11 +1155,11 @@ snapshots:
'@workflow/[email protected]': {}
dependencies:
- '@ai-sdk/gateway': 4.0.85([email protected])
- '@ai-sdk/provider': 4.0.17
- '@ai-sdk/provider-utils': 5.0.43([email protected])
+ '@ai-sdk/gateway': 4.0.104([email protected])
+ '@ai-sdk/provider': 4.0.22
+ '@ai-sdk/provider-utils': 5.0.54([email protected])
zod: 4.4.3
@@ -1219,9 +1219,9 @@ snapshots:
httpxy: 0.5.5
srvx: 1.0.4
dependencies:
- ai: 7.0.105([email protected])
+ ai: 7.0.128([email protected])
nitro: 3.0.260903-beta
undici: 8.9.0
diff --git a/apps/templates/eve-slack-agent-template/package.json b/apps/templates/eve-slack-agent-template/package.json
index fd54d72cb..adfa211e8 100644
--- a/apps/templates/eve-slack-agent-template/package.json
+++ b/apps/templates/eve-slack-agent-template/package.json
@@ -14,7 +14,7 @@
},
"dependencies": {
"@vercel/connect": "^2.4.0",
- "ai": "7.0.105",
+ "ai": "7.0.128",
"eve": "^0.66.3",
"zod": "4.4.3"
},
diff --git a/apps/templates/eve-slack-agent-template/pnpm-lock.yaml b/apps/templates/eve-slack-agent-template/pnpm-lock.yaml
index 43952e985..89c4d2328 100644
--- a/apps/templates/eve-slack-agent-template/pnpm-lock.yaml
+++ b/apps/templates/eve-slack-agent-template/pnpm-lock.yaml
@@ -10,13 +10,13 @@ importers:
dependencies:
'@vercel/connect':
specifier: ^2.4.0
ai:
- specifier: 7.0.105
- version: 7.0.105([email protected])
+ specifier: 7.0.128
+ version: 7.0.128([email protected])
eve:
specifier: ^0.66.3
zod:
specifier: 4.4.3
version: 4.4.3
@@ -30,20 +30,20 @@ importers:
packages:
- '@ai-sdk/[email protected]':
- resolution: {integrity: sha512-c8ztST/CslupqO3p7izsAuG3TlKg9mhE6mGJ42iUADsSMJIV5IOKd9AsLG33QCAY7SWs6NDO2giM/rq7bUTNlQ==}
+ '@ai-sdk/[email protected]':
+ resolution: {integrity: sha512-4lTyt8a4BMfVz4QGo+VO5odakmYMX5kq+xiRgdQGGY7o/jP2WSasp6LvIvYvt/lHzz1MfE+xKTOJhz7mMwjynA==}
engines: {node: '>=22'}
peerDependencies:
zod: ^3.25.76 || ^4.1.8
- '@ai-sdk/[email protected]':
- resolution: {integrity: sha512-gw/bcNseOGSs59TMtV4H1KwqXXe24NHgx+uBYr98pa4Fg6Uvp8hPPxjuckVnfFFyIxHCDew8so0ujpvcSuyMZA==}
+ '@ai-sdk/[email protected]':
+ resolution: {integrity: sha512-amqDxFnw9+dpVrDACRDGmExUUNOH9C1D2bhg/DqWJnamEOieDTXBBXLHujSby/0MqHdAFksnpE8Pzfqx6ABMBw==}
engines: {node: '>=22'}
peerDependencies:
zod: ^3.25.76 || ^4.1.8
- '@ai-sdk/[email protected]':
- resolution: {integrity: sha512-VYMBxIQdcHqbIf1j+YZlI9Ati6LZ4wJe0GGd4z4a5H/KxTggjeOiyaVYTnfF7LHZK5jMQ+rofmzz4QPqf++NUw==}
+ '@ai-sdk/[email protected]':
+ resolution: {integrity: sha512-1Jtmn36VNMOBo9CMUQrPqsR744qtqRloEiXpWCYS8ffPlhS2CrSNmoho5lPjp3R7EnkzBsGI6p51TA7fQJQZdA==}
engines: {node: '>=22'}
'@oxc-project/[email protected]':
@@ -318,8 +318,8 @@ packages:
'@workflow/[email protected]':
resolution: {integrity: sha512-pav4F2BoirECWR7Nf1TKt+2eETcBj7jj4cBefQ8VXQCA6NPkaKeLfj/zMgi+3zYV5ZIBT4GuUiphsj0/b9hPQQ==}
- resolution: {integrity: sha512-gkV3W+tTtWwQ41KNV/KjV8wabBrk+bnbnZ3f0AWp/T9LLyljose8uhFViKleVy7tIdEiR2PiS4BpbNdcExf7rQ==}
+ resolution: {integrity: sha512-WMKN83DodYtH52ETPHq0jrSwj3wU0aHkhLHXMrwMRabC/aMpIqeWsXJgjkUwrOqkaKHWONdupk/2ttPRvAyFcg==}
engines: {node: '>=22'}
peerDependencies:
zod: ^3.25.76 || ^4.1.8
@@ -532,23 +532,23 @@ packages:
snapshots:
dependencies:
- '@ai-sdk/provider': 4.0.17
- '@ai-sdk/provider-utils': 5.0.43([email protected])
+ '@ai-sdk/provider': 4.0.22
+ '@ai-sdk/provider-utils': 5.0.54([email protected])
'@vercel/oidc': 3.2.0
zod: 4.4.3
dependencies:
- '@ai-sdk/provider': 4.0.17
+ '@ai-sdk/provider': 4.0.22
'@standard-schema/spec': 1.1.0
'@workflow/serde': 4.1.0
eventsource-parser: 3.1.1
undici: 7.29.0
zod: 4.4.3
- '@ai-sdk/[email protected]':
+ '@ai-sdk/[email protected]':
dependencies:
json-schema: 0.4.0
@@ -676,12 +676,12 @@ snapshots:
dependencies:
execa: 5.1.1
dependencies:
'@vercel/oidc': 4.0.0
optionalDependencies:
- ai: 7.0.105([email protected])
+ ai: 7.0.128([email protected])
'@vercel/[email protected]': {}
@@ -693,11 +693,11 @@ snapshots:
'@workflow/[email protected]': {}
dependencies:
- '@ai-sdk/gateway': 4.0.85([email protected])
- '@ai-sdk/provider': 4.0.17
- '@ai-sdk/provider-utils': 5.0.43([email protected])
+ '@ai-sdk/gateway': 4.0.104([email protected])
+ '@ai-sdk/provider': 4.0.22
+ '@ai-sdk/provider-utils': 5.0.54([email protected])
zod: 4.4.3
@@ -721,9 +721,9 @@ snapshots:
httpxy: 0.5.5
srvx: 1.0.4
dependencies:
- ai: 7.0.105([email protected])
+ ai: 7.0.128([email protected])
nitro: 3.0.260903-beta
undici: 8.9.0
diff --git a/apps/templates/personal-agent-template/package.json b/apps/templates/personal-agent-template/package.json
index 54a4134c3..ed7e96bc0 100644
--- a/apps/templates/personal-agent-template/package.json
+++ b/apps/templates/personal-agent-template/package.json
@@ -37,7 +37,7 @@
"@shikijs/langs": "^4.4.3",
"@vercel/analytics": "^2.0.1",
"@vercel/connect": "^2.4.0",
- "ai": "7.0.105",
+ "ai": "7.0.128",
"better-auth": "^1.7.1",
"chat": "^4.38.1",
"chat-adapter-sendblue": "^0.2.0",
diff --git a/apps/templates/personal-agent-template/pnpm-lock.yaml b/apps/templates/personal-agent-template/pnpm-lock.yaml
index 49cd22772..14315d60e 100644
--- a/apps/templates/personal-agent-template/pnpm-lock.yaml
+++ b/apps/templates/personal-agent-template/pnpm-lock.yaml
@@ -13,13 +13,13 @@ importers:
'@github-tools/sdk':
specifier: ^1.11.1
'@libsql/client':
specifier: ^0.17.4
version: 0.17.4
'@nuxt/ui':
specifier: ^4.11.1
- version: 4.11.1(bb980e1ad3ed1800d848c88676a3cc2c)
+ version: 4.11.1(3805a5bf82acd8278d90575bde40015f)
'@nuxthub/core':
specifier: ^0.10.8
@@ -31,19 +31,19 @@ importers:
version: 2.0.1([email protected](99e75a557cede4926b71c8c93d703d06))([email protected])([email protected]([email protected]))
'@vercel/connect':
specifier: ^2.4.0
ai:
- specifier: 7.0.105
- version: 7.0.105([email protected])
+ specifier: 7.0.128
+ version: 7.0.128([email protected])
better-auth:
specifier: ^1.7.1
chat:
specifier: ^4.38.1
chat-adapter-sendblue:
specifier: ^0.2.0
drizzle-kit:
specifier: ^0.31.10
version: 0.31.10
@@ -52,7 +52,7 @@ importers:
eve:
specifier: ^0.66.3
maska:
specifier: ^3.2.0
version: 3.2.0
@@ -96,20 +96,20 @@ importers:
packages:
- '@ai-sdk/[email protected]':
- resolution: {integrity: sha512-c8ztST/CslupqO3p7izsAuG3TlKg9mhE6mGJ42iUADsSMJIV5IOKd9AsLG33QCAY7SWs6NDO2giM/rq7bUTNlQ==}
+ '@ai-sdk/[email protected]':
+ resolution: {integrity: sha512-4lTyt8a4BMfVz4QGo+VO5odakmYMX5kq+xiRgdQGGY7o/jP2WSasp6LvIvYvt/lHzz1MfE+xKTOJhz7mMwjynA==}
engines: {node: '>=22'}
peerDependencies:
zod: ^3.25.76 || ^4.1.8
- '@ai-sdk/[email protected]':
- resolution: {integrity: sha512-gw/bcNseOGSs59TMtV4H1KwqXXe24NHgx+uBYr98pa4Fg6Uvp8hPPxjuckVnfFFyIxHCDew8so0ujpvcSuyMZA==}
+ '@ai-sdk/[email protected]':
+ resolution: {integrity: sha512-amqDxFnw9+dpVrDACRDGmExUUNOH9C1D2bhg/DqWJnamEOieDTXBBXLHujSby/0MqHdAFksnpE8Pzfqx6ABMBw==}
engines: {node: '>=22'}
peerDependencies:
zod: ^3.25.76 || ^4.1.8
- '@ai-sdk/[email protected]':
- resolution: {integrity: sha512-VYMBxIQdcHqbIf1j+YZlI9Ati6LZ4wJe0GGd4z4a5H/KxTggjeOiyaVYTnfF7LHZK5jMQ+rofmzz4QPqf++NUw==}
+ '@ai-sdk/[email protected]':
+ resolution: {integrity: sha512-1Jtmn36VNMOBo9CMUQrPqsR744qtqRloEiXpWCYS8ffPlhS2CrSNmoho5lPjp3R7EnkzBsGI6p51TA7fQJQZdA==}
engines: {node: '>=22'}
@@ -3053,8 +3053,8 @@ packages:
resolution: {integrity: sha512-MnA+YT8fwfJPgBx3m60MNqakm30XOkyIoH1y6huTQvC0PwZG7ki8NacLBcrPbNoo8vEZy7Jpuk7+jMO+CUovTQ==}
engines: {node: '>= 14'}
- resolution: {integrity: sha512-gkV3W+tTtWwQ41KNV/KjV8wabBrk+bnbnZ3f0AWp/T9LLyljose8uhFViKleVy7tIdEiR2PiS4BpbNdcExf7rQ==}
+ resolution: {integrity: sha512-WMKN83DodYtH52ETPHq0jrSwj3wU0aHkhLHXMrwMRabC/aMpIqeWsXJgjkUwrOqkaKHWONdupk/2ttPRvAyFcg==}
engines: {node: '>=22'}
peerDependencies:
zod: ^3.25.76 || ^4.1.8
@@ -6532,23 +6532,23 @@ packages:
snapshots:
dependencies:
- '@ai-sdk/provider': 4.0.17
- '@ai-sdk/provider-utils': 5.0.43([email protected])
+ '@ai-sdk/provider': 4.0.22
+ '@ai-sdk/provider-utils': 5.0.54([email protected])
'@vercel/oidc': 3.2.0
zod: 4.4.3
dependencies:
- '@ai-sdk/provider': 4.0.17
+ '@ai-sdk/provider': 4.0.22
'@standard-schema/spec': 1.1.0
'@workflow/serde': 4.1.0
eventsource-parser: 3.1.1
undici: 7.29.0
zod: 4.4.3
- '@ai-sdk/[email protected]':
+ '@ai-sdk/[email protected]':
dependencies:
json-schema: 0.4.0
@@ -7235,14 +7235,14 @@ snapshots:
- '@vue/composition-api'
- vue
dependencies:
- ai: 7.0.105([email protected])
+ ai: 7.0.128([email protected])
octokit: 5.0.5
zod: 4.4.3
optionalDependencies:
'@iconify-json/[email protected]':
dependencies:
@@ -7817,7 +7817,7 @@ snapshots:
rc9: 3.0.1
std-env: 4.2.0
- '@nuxt/[email protected](bb980e1ad3ed1800d848c88676a3cc2c)':
+ '@nuxt/[email protected](3805a5bf82acd8278d90575bde40015f)':
dependencies:
'@floating-ui/dom': 1.8.0
'@iconify/vue': 5.0.1([email protected]([email protected]))
@@ -7886,7 +7886,7 @@ snapshots:
optionalDependencies:
'@internationalized/date': 3.12.3
'@internationalized/number': 3.6.7
- ai: 7.0.105([email protected])
+ ai: 7.0.128([email protected])
zod: 4.4.3
transitivePeerDependencies:
- '@azure/app-configuration'
@@ -9261,13 +9261,13 @@ snapshots:
dependencies:
execa: 5.1.1
dependencies:
'@vercel/oidc': 4.0.0
optionalDependencies:
- ai: 7.0.105([email protected])
+ ai: 7.0.128([email protected])
dependencies:
@@ -9508,11 +9508,11 @@ snapshots:
dependencies:
- '@ai-sdk/gateway': 4.0.85([email protected])
- '@ai-sdk/provider': 4.0.17
- '@ai-sdk/provider-utils': 5.0.43([email protected])
+ '@ai-sdk/gateway': 4.0.104([email protected])
+ '@ai-sdk/provider': 4.0.22
+ '@ai-sdk/provider-utils': 5.0.54([email protected])
zod: 4.4.3
@@ -9770,12 +9770,12 @@ snapshots:
dependencies:
sendblue: 3.11.0
dependencies:
'@workflow/serde': 4.1.0-beta.2
mdast-util-to-string: 4.0.0
@@ -9785,7 +9785,7 @@ snapshots:
remend: 1.3.0
unified: 11.0.5
optionalDependencies:
- ai: 7.0.105([email protected])
+ ai: 7.0.128([email protected])
zod: 4.4.3
transitivePeerDependencies:
- supports-color
@@ -10296,13 +10296,13 @@ snapshots:
dependencies:
- ai: 7.0.105([email protected])
+ ai: 7.0.128([email protected])
nitro: 3.0.260903-beta
undici: 8.9.0
optionalDependencies:
microsandbox: 0.6.12
diff --git a/docs/agent-config.md b/docs/agent-config.md
index 5161f764f..bb3441d94 100644
--- a/docs/agent-config.md
+++ b/docs/agent-config.md
@@ -69,8 +69,8 @@ takes precedence; AI Gateway hashes IDs longer than 256 characters.
### Choose the model dynamically
-To select a model from the incoming prompt with an AI SDK evaluation model, use
-[`auto` from `eve/models`](./guides/evaluate).
+To select a model from the incoming prompt with an AI SDK decision model, use
+[`auto` from `eve/models`](./guides/decide).
`model` also accepts `defineDynamic({ events })`. Each matching handler must
return the concrete model for its scope; a dynamic model has no compiled
diff --git a/docs/connections/overview.mdx b/docs/connections/overview.mdx
index 3ce7085af..6e55fe896 100644
--- a/docs/connections/overview.mdx
+++ b/docs/connections/overview.mdx
@@ -194,7 +194,7 @@ export default defineMcpClientConnection({
});
```
-`never()` lets every call through, `once()` asks for approval the first time in a session, `always()` asks every time, and `auto()` uses an evaluation model to allow routine calls and ask about dangerous or unclear effects. The pause and resume is the same human-in-the-loop flow covered in [Tools](/docs/tools).
+`never()` lets every call through, `once()` asks for approval the first time in a session, `always()` asks every time, and `auto()` uses a decision model to allow routine calls and ask about dangerous or unclear effects. The pause and resume is the same human-in-the-loop flow covered in [Tools](/docs/tools).
For connection tools that can create, modify, delete, transmit, purchase, message, or access sensitive data, use approval, allow-lists, or other safeguards appropriate to the action.
diff --git a/docs/dynamic-configuration/meta.json b/docs/dynamic-configuration/meta.json
index 998c9d957..9c7a2c824 100644
--- a/docs/dynamic-configuration/meta.json
+++ b/docs/dynamic-configuration/meta.json
@@ -1,5 +1,5 @@
{
"title": "Dynamic Capabilities",
"defaultOpen": false,
- "pages": ["../guides/dynamic-capabilities", "../guides/evaluate"]
+ "pages": ["../guides/dynamic-capabilities", "../guides/decide"]
}
diff --git a/docs/evals/judge.mdx b/docs/evals/judge.mdx
index 62fd23667..2b039f3c1 100644
--- a/docs/evals/judge.mdx
+++ b/docs/evals/judge.mdx
@@ -1,9 +1,9 @@
---
title: "Judge"
-description: "Grade evals with evaluation models using criteria, typed questions, or batches, and set thresholds on each assertion."
+description: "Grade evals with decision models using criteria, typed questions, or batches, and set thresholds on each assertion."
---
-Use `t.judge(...)` when a deterministic [assertion](./assertions) cannot capture what a good response looks like. It calls [`evaluate` from `eve/ai`](../guides/evaluate#evaluate-inside-a-tool) and defaults to `typesafe-ai/jev`, TypeSafe AI's [Jev evaluation model](https://vercel.com/i/what-is-jev). The judge is separate from the agent under test and is used only for scoring.
+Use `t.judge(...)` when a deterministic [assertion](./assertions) cannot capture what a good response looks like. It calls [`decide` from `eve/ai`](../guides/decide#decide-inside-a-tool) and defaults to `typesafe-ai/jev`, TypeSafe AI's [Jev decision model](https://vercel.com/i/what-is-jev). The judge is separate from the agent under test and is used only for scoring.
```ts
import { defineEval } from "eve/evals";
@@ -17,7 +17,7 @@ export default defineEval({
});
```
-No judge configuration is required. Configure credentials for the evaluator as described under [Configuring the judge model](#configuring-the-judge-model).
+No judge configuration is required. Configure credentials for the decider as described under [Configuring the judge model](#configuring-the-judge-model).
## The graders
@@ -41,11 +41,11 @@ t.judge({
.atLeast(0.75);
```
-Choice questions require an `expected` option key. eve uses that expectation locally for scoring; it does not send it to the evaluator. Instructions and rubric descriptions accept strings, JSON objects, or JSON arrays. Descriptions can also be `null`.
+Choice questions require an `expected` option key. eve uses that expectation locally for scoring; it does not send it to the decider. Instructions and rubric descriptions accept strings, JSON objects, or JSON arrays. Descriptions can also be `null`.
## Choose the value to grade
-By default, the evaluator receives `{ input, output }`: the latest prompt and assistant message from the session whose turn most recently settled. Approval responses and stream reads preserve that prompt. For multipart messages, including `session.sendFile()`, the prompt contains the text parts.
+By default, the decider receives `{ input, output }`: the latest prompt and assistant message from the session whose turn most recently settled. Approval responses and stream reads preserve that prompt. For multipart messages, including `session.sendFile()`, the prompt contains the text parts.
Pass `on` to replace the output. JSON values stay structured rather than being converted to strings:
@@ -66,9 +66,9 @@ t.judge("The assistant remembers the user's favorite word across turns.", {
The transcript contains user and assistant messages in turn order, excluding reasoning, tool calls, and tool results. See [Multi-turn evals](./cases#multi-turn-evals).
-## Evaluate several questions together
+## Decide several questions together
-A batch makes one evaluation request against one shared state. Pass `state` to replace the default `{ input, output }` completely. Batch calls use `state` instead of `on`.
+A batch makes one decision request against one shared state. Pass `state` to replace the default `{ input, output }` completely. Batch calls use `state` instead of `on`.
```ts
const judgments = t.judge({
@@ -120,9 +120,9 @@ Model settings resolve in this order:
1. Per-call: `t.judge("…", { model, modelOptions })`.
2. Per-eval: `defineEval({ judge: { model, modelOptions }, test })`.
3. Project default: `defineEvalConfig({ judge: { model, modelOptions } })`.
-4. The default model used by `evaluate`, currently `typesafe-ai/jev`.
+4. The default model used by `decide`, currently `typesafe-ai/jev`.
-A per-eval `judge` replaces the project's judge configuration. Per-call fields override the resolved configuration. `judge` and its `model` are optional; deterministic evals make no evaluation requests.
+A per-eval `judge` replaces the project's judge configuration. Per-call fields override the resolved configuration. `judge` and its `model` are optional; deterministic evals make no decision requests.
```ts title="evals/evals.config.ts"
import { defineEvalConfig } from "eve/evals";
@@ -132,30 +132,30 @@ export default defineEvalConfig({
});
```
-String IDs resolve through the AI SDK default provider, or Vercel AI Gateway when none is configured. Gateway uses `AI_GATEWAY_API_KEY` or Vercel OIDC credentials; eve's existing local Gateway connection is also honored by `evaluate`. An explicit provider evaluation model uses that provider's credentials:
+String IDs resolve through the AI SDK default provider, or Vercel AI Gateway when none is configured. Gateway uses `AI_GATEWAY_API_KEY` or Vercel OIDC credentials; eve's existing local Gateway connection is also honored by `decide`. An explicit provider decision model uses that provider's credentials:
```ts
import { openai } from "@ai-sdk/openai";
t.judge("The response addresses the request.", {
- model: openai.evaluationModel("gpt-6-luna"),
+ model: openai.decisionModel("gpt-6-luna"),
modelOptions: { providerOptions: { openai: { reasoningEffort: "high" } } },
});
```
-Use the provider's `evaluationModel` factory, not a language model instance. AI SDK supplies evaluation adapters for supported language models. Provider support and the evaluation contract are experimental; see [AI SDK evaluation](https://ai-sdk.dev/docs/ai-sdk-core/evaluation).
+Use the provider's `decisionModel` factory, not a language model instance. AI SDK supplies decision adapters for supported language models. Provider support and the decision contract are experimental; see [AI SDK decision](https://ai-sdk.dev/docs/ai-sdk-core/decisions).
-String model IDs resolve through the default provider's evaluation API. Gateway accepts native evaluation models such as `typesafe-ai/jev`; passing a language model ID such as `openai/gpt-6-luna` does not automatically select the OpenAI adapter. Use an evaluation-model instance as shown above for language models.
+String model IDs resolve through the default provider's decision API. Gateway accepts native decision models such as `typesafe-ai/jev`; passing a language model ID such as `openai/gpt-6-luna` does not automatically select the OpenAI adapter. Use a decision-model instance as shown above for language models.
-Missing credentials, invalid answers, unsupported question types, and provider errors become failed gates even for tracked-only judgments. They do not silently skip the eval or select another model. Validation, authentication, and retries use the shared `evaluate` implementation.
+Missing credentials, invalid answers, unsupported question types, and provider errors become failed gates even for tracked-only judgments. They do not silently skip the eval or select another model. Validation, authentication, and retries use the shared `decide` implementation.
## Diagnostics and migration
-Reports use `judge.boolean`, `judge.score`, and `judge.choice`; batch assertion names also include their question key. Diagnostics include the captured state, question, and raw answer. Assertion metadata retains the normalized score, model identity, token usage, warnings, and available provider and response metadata. Usage for a batch is marked as shared across its assertions, not a separate charge for each question. Evaluation models do not promise a prose rationale.
+Reports use `judge.boolean`, `judge.score`, and `judge.choice`; batch assertion names also include their question key. Diagnostics include the captured state, question, and raw answer. Assertion metadata retains the normalized score, model identity, token usage, warnings, and available provider and response metadata. Usage for a batch is marked as shared across its assertions, not a separate charge for each question. Decision models do not promise a prose rationale.
The autoevals graders have been removed. Replace `t.judge.autoevals.closedQA(criteria, options)` with `t.judge(criteria, options)`. Scores now represent a probability rather than a binary yes/no answer, so review thresholds against your examples. For factuality, summaries, or SQL equivalence, write an explicit question and include the reference in shared state. No old grader buckets are preserved.
-Change configured language model instances to evaluation model instances. `modelOptions.providerOptions`, assertion labels, and threshold methods remain available. [Braintrust reporting](./reporters) continues to accept judge scores independently of the grader implementation.
+Change configured language model instances to decision model instances. `modelOptions.providerOptions`, assertion labels, and threshold methods remain available. [Braintrust reporting](./reporters) continues to accept judge scores independently of the grader implementation.
## What to read next
diff --git a/docs/evals/overview.mdx b/docs/evals/overview.mdx
index 8a32a0dd1..f23fd825e 100644
--- a/docs/evals/overview.mdx
+++ b/docs/evals/overview.mdx
@@ -56,7 +56,7 @@ export default defineEvalConfig({
});
```
-Everything is optional. `judge` sets the default model for [LLM-as-judge](./judge) assertions (`t.judge(...)`); omit it to use the default evaluator, `typesafe-ai/jev`. Deterministic evals make no judge requests. `reporters`, `maxConcurrency`, and `timeoutMs` round out the defaults. Config `reporters` observe every eval in the run, so set one `Braintrust()` here instead of adding it to each eval. CLI flags (`--max-concurrency`, `--timeout`) and per-eval values take precedence over the config defaults.
+Everything is optional. `judge` sets the default model for [LLM-as-judge](./judge) assertions (`t.judge(...)`); omit it to use the default decider, `typesafe-ai/jev`. Deterministic evals make no judge requests. `reporters`, `maxConcurrency`, and `timeoutMs` round out the defaults. Config `reporters` observe every eval in the run, so set one `Braintrust()` here instead of adding it to each eval. CLI flags (`--max-concurrency`, `--timeout`) and per-eval values take precedence over the config defaults.
### Run setup and teardown
diff --git a/docs/guides/evaluate.md b/docs/guides/decide.md
similarity index 68%
rename from docs/guides/evaluate.md
rename to docs/guides/decide.md
index b75167e01..6bf9eb9bc 100644
--- a/docs/guides/evaluate.md
+++ b/docs/guides/decide.md
@@ -1,21 +1,21 @@
---
title: Automatic Model Selection
-description: "Choose agent models automatically or evaluate typed questions in your tools and application code."
+description: "Choose agent models automatically or ask typed questions in your tools and application code."
---
Use `auto` from `eve/models` to choose an agent model from an allowlist before
-inference begins. It uses the [AI SDK evaluation API](https://ai-sdk.dev/docs/ai-sdk-core/evaluation),
-so the evaluator can be a Vercel AI Gateway model ID or an evaluation model from
-an installed provider. Use `evaluate` from `eve/ai` to ask typed questions in
+inference begins. It uses the [AI SDK decision API](https://ai-sdk.dev/docs/ai-sdk-core/decisions),
+so the decider can be a Vercel AI Gateway model ID or a decision model from
+an installed provider. Use `decide` from `eve/ai` to ask typed questions in
your own tools or application code.
-The AI SDK evaluation model specification is experimental and can change in
+The AI SDK decision model specification is experimental and can change in
patch releases.
## Choose from Gateway models
-By default, `auto` evaluates with `typesafe-ai/jev`, TypeSafe AI's
-[Jev evaluation model](https://vercel.com/i/what-is-jev). Like other AI SDK
+By default, `auto` decides with `typesafe-ai/jev`, TypeSafe AI's
+[Jev decision model](https://vercel.com/i/what-is-jev). Like other AI SDK
model strings, it uses Vercel AI Gateway unless the application has configured a
different global default provider.
@@ -35,13 +35,13 @@ export default defineAgent({
Configure Gateway authentication as you would for any other AI SDK model. eve
does not add a TypeSafe credential or transport layer. During `eve dev`, a
-Gateway evaluator uses the same connection selected through `/login` as Gateway
+Gateway decider uses the same connection selected through `/login` as Gateway
language models. A configured AI SDK default provider still owns string model
resolution during development. The TUI footer displays `dynamic model` when the
agent uses `auto`, then adds the resolved model for the current turn, such as
`dynamic model · openai/gpt-6-luna`.
-Set `fallback` to keep the turn running when the evaluator is unavailable or
+Set `fallback` to keep the turn running when the decider is unavailable or
returns an error. The fallback can be a Gateway model ID, an AI SDK language
model instance, or an object with `model` and a `reasoning` override:
@@ -60,15 +60,15 @@ export default defineAgent({
});
```
-The fallback applies only to evaluator failures. eve logs a warning with the
-evaluator error and fallback identity when it switches models. Invalid prompts,
+The fallback applies only to decider failures. eve logs a warning with the
+decider error and fallback identity when it switches models. Invalid prompts,
including requests without user text or with a latest message over the routing
-limit, still fail before evaluator I/O. Cancelling the turn also aborts routing
+limit, still fail before decider I/O. Cancelling the turn also aborts routing
instead of selecting the fallback.
## Use a provider directly
-Install the provider package yourself and pass its evaluation model. The provider
+Install the provider package yourself and pass its decision model. The provider
owns its credentials and settings.
```sh
@@ -82,7 +82,7 @@ import { auto } from "eve/models";
export default defineAgent({
model: auto({
- model: typeSafeAi.evaluationModel("jev-latest"),
+ model: typeSafeAi.decisionModel("jev-latest"),
options: {
"openai/gpt-6-sol": "Difficult reasoning and engineering tasks",
"openai/gpt-6-luna": "Routine tasks where fast completion matters",
@@ -91,12 +91,12 @@ export default defineAgent({
});
```
-Any provider that implements the AI SDK `Experimental_EvaluationModel` contract
+Any provider that implements the AI SDK `Experimental_DecisionModel` contract
works here.
## Route to provider models and set reasoning
-An option's key is the value shown to the evaluator. A string value describes a
+An option's key is the value shown to the decider. A string value describes a
Gateway model whose ID is the key. Use the object form when the selected model is
a provider instance, an alias, or needs a reasoning override.
@@ -120,7 +120,7 @@ export default defineAgent({
});
```
-The evaluator sees option keys, descriptions, and recent text messages. It never
+The decider sees option keys, descriptions, and recent text messages. It never
receives provider credentials or serialized language model instances. When it
selects `my_secret_model`, eve resolves the key back to the authored Anthropic
model.
@@ -129,16 +129,16 @@ Supported reasoning values are `"provider-default"`, `"none"`, `"minimal"`,
`"low"`, `"medium"`, `"high"`, and `"xhigh"`. An omitted value inherits the
agent's reasoning setting.
-## Evaluate inside a tool
+## Decide inside a tool
-Use `evaluate` to ask choice, score, or boolean questions about the state you pass to it.
+Use `decide` to ask choice, score, or boolean questions about the state you pass to it.
It defaults to `typesafe-ai/jev` and uses the same authentication as `auto`,
including the Gateway connection selected through `/login` during `eve dev`.
-Pass `model` to use another evaluation model ID or a provider instance. A configured
+Pass `model` to use another decision model ID or a provider instance. A configured
AI SDK default provider takes precedence over the local Gateway connection.
```ts title="agent/tools/classify-request.ts"
-import { evaluate } from "eve/ai";
+import { decide } from "eve/ai";
import { defineTool } from "eve/tools";
import { z } from "zod";
@@ -146,7 +146,7 @@ export default defineTool({
description: "Choose the team that can help with a customer request.",
inputSchema: z.object({ request: z.string().min(1).max(8000) }),
async execute({ request }, ctx) {
- const result = await evaluate({
+ const result = await decide({
state: { request },
questions: {
team: {
@@ -169,23 +169,23 @@ The choice above is typed as `"billing" | "support"`. Each question appears unde
its authored key in `result.answers`. Results also include token usage, warnings,
provider metadata, and response metadata. To use that choice to delegate while keeping specialist subagents out of the parent model's tools, see [Route to a hidden subagent with Jev](/docs/tools/workflows#route-to-a-hidden-subagent-with-jev).
-`evaluate` accepts AI SDK evaluation options, including `maxRetries`, `headers`,
+`decide` accepts AI SDK decision options, including `maxRetries`, `headers`,
and `providerOptions`. Pass an `abortSignal` to cancel the request. Input and
answer validation, retries, and provider errors follow AI SDK semantics.
-You can also call `evaluate` outside a tool; it does not require an active eve
-session. Each call performs its own evaluation. `auto` uses this function
+You can also call `decide` outside a tool; it does not require an active eve
+session. Each call performs its own decision. `auto` uses this function
and adds the per-turn routing behavior described below.
## Judge eval results
-Eval authors can use `t.judge(...)` to turn evaluation answers into scored assertions, including batches of questions sharing one state. It uses this same `evaluate` implementation and default model. See [Judge](../evals/judge) for criteria, rubrics, and thresholds.
+Eval authors can use `t.judge(...)` to turn decision answers into scored assertions, including batches of questions sharing one state. It uses this same `decide` implementation and default model. See [Judge](../evals/judge) for criteria, rubrics, and thresholds.
-## Evaluate tool approvals
+## Decide tool approvals
-Use `auto({ model? })` when an evaluation model should decide whether a
+Use `auto({ model? })` when a decision model should decide whether a
tool call can run automatically or needs human approval. It accepts the same
-AI SDK evaluation model strings and provider instances described above and
+AI SDK decision model strings and provider instances described above and
defaults to `typesafe-ai/jev`:
```ts title="agent/tools/deploy.ts"
@@ -201,7 +201,7 @@ export default defineTool({
});
```
-The evaluation model reviews the tool name and input for dangerous effects. A
+The decision model reviews the tool name and input for dangerous effects. A
caution, failed review, or incomplete input requires human approval. See
[Human-in-the-loop approvals](/docs/human-in-the-loop#approvals) for classifier
options and data handling, and the
@@ -210,17 +210,17 @@ guide for an end-to-end walkthrough.
## Runtime behavior
-`auto` evaluates at the first `step.started` event, after the incoming prompt
+`auto` decides at the first `step.started` event, after the incoming prompt
is available and before the selected language model runs. It reuses that choice
for later tool-loop steps in the same turn. A new turn makes a new choice, and
child sessions route from their own prompts.
-The evaluator receives up to eight recent user and assistant text messages,
+The decider receives up to eight recent user and assistant text messages,
capped at 16,000 characters. Requests without user text and latest messages over
the limit fail before provider I/O.
-Evaluation validation, retries, provider errors, and model resolution follow AI
-SDK semantics. Without `fallback`, an evaluator error fails the turn. With
+Decision validation, retries, provider errors, and model resolution follow AI
+SDK semantics. Without `fallback`, a decider error fails the turn. With
`fallback`, eve uses and retains that model for the rest of the turn. Cancelling
-the active turn aborts evaluation and prevents a routed or fallback choice from
+the active turn aborts decision and prevents a routed or fallback choice from
being retained.
diff --git a/docs/reference/typescript-api.md b/docs/reference/typescript-api.md
index d09fe1848..c5798373c 100644
--- a/docs/reference/typescript-api.md
+++ b/docs/reference/typescript-api.md
@@ -143,10 +143,10 @@ Tool definitions accept `availableInSubagents: false` to restrict the tool to to
| `eve/instrumentation` | `defineInstrumentation`, `disableInstrumentation`, `isChannel`, lifecycle provider types |
| `eve/instrumentation/otel` | `otel`, `otelIntegration`, `localTraces`, `agentRuns`, OpenTelemetry policy types |
| `eve/local-dev` | `getLocalDevCapability`, `LocalDevCapability` |
-| [`eve/models`](../guides/evaluate) | Automatic model selection with `auto` |
+| [`eve/models`](../guides/decide) | Automatic model selection with `auto` |
| `eve/models/openai` | `openai`, `chatgpt`, deprecated `experimental_chatgpt` |
| `eve/models/anthropic` | `anthropic` |
-| [`eve/ai`](../guides/evaluate#evaluate-inside-a-tool) | Standalone `evaluate` |
+| [`eve/ai`](../guides/decide#decide-inside-a-tool) | Standalone `decide` |
| `eve/evals` | `defineEval`, `defineEvalConfig`, `mockModel`, eval types |
| `eve/evals/expect` | `includes`, `equals`, `matches`, `similarity` |
| `eve/evals/reporters` | `Braintrust`, `JUnit`, `EvalReporter` |
diff --git a/docs/tools/human-in-the-loop.md b/docs/tools/human-in-the-loop.md
index dacb6ffaf..d38ca556d 100644
--- a/docs/tools/human-in-the-loop.md
+++ b/docs/tools/human-in-the-loop.md
@@ -37,17 +37,17 @@ export default defineTool({
| `never()` | Never require approval (the default when omitted). |
| `once()` | Require approval only the first time the tool runs in a session; auto-allow after. |
| `always()` | Require approval before every call. |
-| `auto()` | Ask an evaluation model whether to run the exact call or require user approval. |
+| `auto()` | Ask a decision model whether to run the exact call or require user approval. |
By default, omitted `approval` behaves like `never()`, so tool calls may execute without human approval. Require human approval or other safeguards for sensitive, irreversible, regulated, financial, healthcare, employment, housing, legal, safety-impacting, user-impacting, or external side-effecting actions.
-`auto()` uses an [AI SDK evaluation model](/docs/guides/evaluate) to classify each call as `clear` or `caution`. It defaults to `typesafe-ai/jev`, TypeSafe AI's [Jev evaluation model](https://vercel.com/i/what-is-jev). Like `evaluate`, a model string uses Vercel AI Gateway unless the application configures a global AI SDK default provider:
+`auto()` uses an [AI SDK decision model](/docs/guides/decide) to classify each call as `clear` or `caution`. It defaults to `typesafe-ai/jev`, TypeSafe AI's [Jev decision model](https://vercel.com/i/what-is-jev). Like `decide`, a model string uses Vercel AI Gateway unless the application configures a global AI SDK default provider:
```ts
approval: auto({ model: "typesafe-ai/jev" });
```
-The evaluation model reviews the tool name and input for dangerous effects. A caution, failed review, or incomplete input requires user approval. The tool input is sent to the evaluation model's provider.
+The decision model reviews the tool name and input for dangerous effects. A caution, failed review, or incomplete input requires user approval. The tool input is sent to the decision model's provider.
Override the classifier text for application-specific policy:
diff --git a/docs/tools/workflows.mdx b/docs/tools/workflows.mdx
index d272538fe..5141f885a 100644
--- a/docs/tools/workflows.mdx
+++ b/docs/tools/workflows.mdx
@@ -588,7 +588,7 @@ export default defineAgent({
Define the model-visible router under `agent/tools/`. Build the Jev criteria from the subagents' effective descriptions, then pass its typed choice directly to `ctx.agent()`:
```ts title="agent/tools/agent-router.ts"
-import { evaluate } from "eve/ai";
+import { decide } from "eve/ai";
import { defineWorkflowTool } from "eve/tools";
import { z } from "zod";
@@ -599,7 +599,7 @@ async function chooseTarget(
) {
"use step";
- const result = await evaluate({
+ const result = await decide({
abortSignal,
state: { task },
questions: {
@@ -637,7 +637,7 @@ export default defineWorkflowTool({
});
```
-The evaluation runs in a step so workflow replay records its result instead of making the routing request again. The parent model receives `agent-router`, but not `researcher` or `operator`. Jev returns the typed `"researcher" | "operator"` choice, and the selected subagent's reply becomes the `agent-router` tool result.
+The decision runs in a step so workflow replay records its result instead of making the routing request again. The parent model receives `agent-router`, but not `researcher` or `operator`. Jev returns the typed `"researcher" | "operator"` choice, and the selected subagent's reply becomes the `agent-router` tool result.
`ctx.agents` is a replay-stable metadata snapshot taken when the session admits the call. In a `serve` task, it follows the call being served. In a top-level root workflow, `ctx.agents.agent` always represents the root-copy target and carries the root's authored `description` when provided. The snapshot also includes declared agents hidden from the parent model with `tool: false` or `disableTool()`, but exposes no model definitions, credentials, or callbacks. A delegated root copy omits `agent`, so `agentRouter()` cannot select another root copy recursively. Invocation still checks each target's availability through `ctx.agent()`.
@@ -651,7 +651,7 @@ import { agentRouter } from "eve/tools/agent-router";
export default agentRouter();
```
-Its input is `{ message: string }`. `agentRouter()` ignores entries without a non-empty description. With two or more described targets, it sends the message and effective descriptions to Jev, then sends the message to a new session with the selected agent through `ctx.agent()` and returns its reply. It invokes a sole target without evaluation. Each call
+Its input is `{ message: string }`. `agentRouter()` ignores entries without a non-empty description. With two or more described targets, it sends the message and effective descriptions to Jev, then sends the message to a new session with the selected agent through `ctx.agent()` and returns its reply. It invokes a sole target without decision. Each call
runs as a [task](#run-calls-as-tasks-task), so the reply reaches the model as the task's result. In a top-level root session, the candidates include the root-copy `agent`; delegated root copies and declared subagent contexts include only their declared targets.
When an authored tool accepts a JSON Schema supplied by the model, represent that input with a permissive object such as `z.looseObject({})`. Avoid `z.record(z.string(), z.json())`: its generated JSON Schema uses `propertyNames`, which OpenAI does not support. Validate the supplied value at the point where your tool consumes it.
diff --git a/e2e/README.md b/e2e/README.md
index a8b24d5be..07a379c1f 100644
--- a/e2e/README.md
+++ b/e2e/README.md
@@ -136,7 +136,7 @@ alias env (local matrix, plain `eve eval --strict`) the eval skips.
Most fixture agents resolve `EVE_E2E_MODEL`
through `@eve-e2e/config`, defaulting to `openai/gpt-6-sol` for local runs.
-Fixture eval configs use the shared `e2eJudgeModel()` helper, which returns an OpenAI evaluation-model instance for `openai/gpt-5.6-luna` until CI has access to Jev. The adapter uses Gateway's Responses endpoint and `AI_GATEWAY_API_KEY`, independently of the agent matrix. A bare Luna string targets Gateway's native evaluation API and is not supported. Deterministic judge coverage in `agent-evaluate` passes a fixture evaluation model explicitly.
+Fixture eval configs use the shared `e2eJudgeModel()` helper, which returns an OpenAI decision-model instance for `openai/gpt-5.6-luna` until CI has access to Jev. The adapter uses Gateway's Responses endpoint and `AI_GATEWAY_API_KEY`, independently of the agent matrix. A bare Luna string targets Gateway's native decision API and is not supported. Deterministic judge coverage in `agent-decide` passes a fixture decision model explicitly.
`agent-workflow-stress` uses eve's `mockModel` fixture helper so its 100-turn
runs stay fast and deterministic. Its concurrent and sequential evals cover
diff --git a/e2e/fixtures/agent-evaluate/agent/agent.ts b/e2e/fixtures/agent-decide/agent/agent.ts
similarity index 93%
rename from e2e/fixtures/agent-evaluate/agent/agent.ts
rename to e2e/fixtures/agent-decide/agent/agent.ts
index 48ac09720..ae5eab63d 100644
--- a/e2e/fixtures/agent-evaluate/agent/agent.ts
+++ b/e2e/fixtures/agent-decide/agent/agent.ts
@@ -44,14 +44,14 @@ export default defineAgent({
],
};
}
- if (request.userMessages.some((text) => text.includes("evaluate-request"))) {
- const result = request.toolResults.find((result) => result.name === "evaluate-request");
+ if (request.userMessages.some((text) => text.includes("decide-request"))) {
+ const result = request.toolResults.find((result) => result.name === "decide-request");
if (result) return JSON.stringify({ isError: result.isError, output: result.output });
return {
toolCalls: [
{
- id: "evaluate-request-1",
- name: "evaluate-request",
+ id: "decide-request-1",
+ name: "decide-request",
input: {
missingAnswer: request.userMessages.some((text) => text.includes("missing answer")),
},
diff --git a/e2e/fixtures/agent-evaluate/agent/instructions.md b/e2e/fixtures/agent-decide/agent/instructions.md
similarity index 100%
rename from e2e/fixtures/agent-evaluate/agent/instructions.md
rename to e2e/fixtures/agent-decide/agent/instructions.md
diff --git a/e2e/fixtures/agent-evaluate/agent/subagents/worker/agent.ts b/e2e/fixtures/agent-decide/agent/subagents/worker/agent.ts
similarity index 100%
rename from e2e/fixtures/agent-evaluate/agent/subagents/worker/agent.ts
rename to e2e/fixtures/agent-decide/agent/subagents/worker/agent.ts
diff --git a/e2e/fixtures/agent-evaluate/agent/subagents/worker/instructions.md b/e2e/fixtures/agent-decide/agent/subagents/worker/instructions.md
similarity index 100%
rename from e2e/fixtures/agent-evaluate/agent/subagents/worker/instructions.md
rename to e2e/fixtures/agent-decide/agent/subagents/worker/instructions.md
diff --git a/e2e/fixtures/agent-evaluate/agent/testing.ts b/e2e/fixtures/agent-decide/agent/testing.ts
similarity index 78%
rename from e2e/fixtures/agent-evaluate/agent/testing.ts
rename to e2e/fixtures/agent-decide/agent/testing.ts
index 8d8c16da6..6def83a93 100644
--- a/e2e/fixtures/agent-evaluate/agent/testing.ts
+++ b/e2e/fixtures/agent-decide/agent/testing.ts
@@ -1,21 +1,21 @@
-import type { Experimental_EvaluationModel } from "ai";
+import type { Experimental_DecisionModel } from "ai";
import { defineDynamic } from "eve";
import { defineState } from "eve/context";
import { mockModel, type MockModelResponder } from "eve/evals";
import { auto } from "eve/models";
-export const routing = defineState("evaluate-fixture.routing", () => ({
+export const routing = defineState("decide-fixture.routing", () => ({
requests: 0,
model: "unselected",
reasoning: "unselected",
}));
-export const permissionEvaluationModel: Exclude<Experimental_EvaluationModel, string> = {
+export const permissionDecisionModel: Exclude<Experimental_DecisionModel, string> = {
specificationVersion: "v4",
provider: "fixture",
- modelId: "fixture-permission-evaluator",
+ modelId: "fixture-permission-decider",
supportedQuestionTypes: ["choice"],
- async doEvaluate({ state }) {
+ async doDecide({ state }) {
const choice = JSON.stringify(state).includes('"effect":"malicious"') ? "caution" : "clear";
return {
answers: { permission: { type: "choice", choice } },
@@ -25,16 +25,16 @@ export const permissionEvaluationModel: Exclude<Experimental_EvaluationModel, st
},
};
-export const evaluationModel: Exclude<Experimental_EvaluationModel, string> = {
+export const decisionModel: Exclude<Experimental_DecisionModel, string> = {
specificationVersion: "v4",
provider: "fixture",
- modelId: "fixture-evaluator",
+ modelId: "fixture-decider",
supportedQuestionTypes: ["choice"],
- async doEvaluate({ state, questions }) {
+ async doDecide({ state, questions }) {
routing.update((value) => ({ ...value, requests: value.requests + 1 }));
const serialized = JSON.stringify(state);
if (serialized.includes("service unavailable")) {
- throw new Error("Evaluation service unavailable.");
+ throw new Error("Decision service unavailable.");
}
const choice = serialized.includes("difficult") ? "openai/large" : "openai/small";
return {
@@ -49,15 +49,15 @@ export const evaluationModel: Exclude<Experimental_EvaluationModel, string> = {
},
usage: { inputTokens: 42, outputTokens: 3 },
warnings: [],
- response: { modelId: "fixture-evaluator" },
+ response: { modelId: "fixture-decider" },
};
},
};
-/** Run the real router with deterministic evaluation and language models. */
+/** Run the real router with deterministic decision and language models. */
export function fixtureModel(respond: MockModelResponder) {
const model = auto({
- model: evaluationModel,
+ model: decisionModel,
options: {
"openai/large": {
model: mockModel({ modelId: "openai/large", respond }),
diff --git a/e2e/fixtures/agent-evaluate/agent/tools/ask_question.ts b/e2e/fixtures/agent-decide/agent/tools/ask_question.ts
similarity index 100%
rename from e2e/fixtures/agent-evaluate/agent/tools/ask_question.ts
rename to e2e/fixtures/agent-decide/agent/tools/ask_question.ts
diff --git a/e2e/fixtures/agent-evaluate/agent/tools/automatic-review.ts b/e2e/fixtures/agent-decide/agent/tools/automatic-review.ts
similarity index 68%
rename from e2e/fixtures/agent-evaluate/agent/tools/automatic-review.ts
rename to e2e/fixtures/agent-decide/agent/tools/automatic-review.ts
index 16be5b3a0..4bee865b9 100644
--- a/e2e/fixtures/agent-evaluate/agent/tools/automatic-review.ts
+++ b/e2e/fixtures/agent-decide/agent/tools/automatic-review.ts
@@ -1,17 +1,17 @@
import { defineTool } from "eve/tools";
import { auto } from "eve/tools/approval";
-import { permissionEvaluationModel } from "../testing";
+import { permissionDecisionModel } from "../testing";
export default defineTool({
- description: "Exercise evaluation-model approval.",
+ description: "Exercise decision-model approval.",
inputSchema: {
type: "object",
properties: { effect: { type: "string", enum: ["safe", "malicious"] } },
required: ["effect"],
additionalProperties: false,
},
- approval: auto({ model: permissionEvaluationModel }),
+ approval: auto({ model: permissionDecisionModel }),
execute({ effect }) {
return { effect, executed: true };
},
diff --git a/e2e/fixtures/agent-evaluate/agent/tools/evaluate-request.ts b/e2e/fixtures/agent-decide/agent/tools/decide-request.ts
similarity index 70%
rename from e2e/fixtures/agent-evaluate/agent/tools/evaluate-request.ts
rename to e2e/fixtures/agent-decide/agent/tools/decide-request.ts
index 21f1dc449..e73cf25ab 100644
--- a/e2e/fixtures/agent-evaluate/agent/tools/evaluate-request.ts
+++ b/e2e/fixtures/agent-decide/agent/tools/decide-request.ts
@@ -1,10 +1,10 @@
-import { evaluate } from "eve/ai";
+import { decide } from "eve/ai";
import { defineTool } from "eve/tools";
-import { evaluationModel } from "../testing";
+import { decisionModel } from "../testing";
export default defineTool({
- description: "Evaluate Alice's request with the fixture evaluation provider.",
+ description: "Decide Alice's request with the fixture decision provider.",
inputSchema: {
type: "object",
properties: { missingAnswer: { type: "boolean" } },
@@ -12,10 +12,10 @@ export default defineTool({
additionalProperties: false,
},
async execute({ missingAnswer }, ctx) {
- const result = await evaluate({
+ const result = await decide({
model: missingAnswer
- ? { ...evaluationModel, doEvaluate: async () => ({ answers: {}, warnings: [] }) }
- : evaluationModel,
+ ? { ...decisionModel, doDecide: async () => ({ answers: {}, warnings: [] }) }
+ : decisionModel,
state: { request: "Alice needs a routine summary." },
questions: {
route: {
diff --git a/e2e/fixtures/agent-evaluate/evals/auto-resume-question.eval.ts b/e2e/fixtures/agent-decide/evals/auto-resume-question.eval.ts
similarity index 100%
rename from e2e/fixtures/agent-evaluate/evals/auto-resume-question.eval.ts
rename to e2e/fixtures/agent-decide/evals/auto-resume-question.eval.ts
diff --git a/e2e/fixtures/agent-evaluate/evals/automatic-approval.eval.ts b/e2e/fixtures/agent-decide/evals/automatic-approval.eval.ts
similarity index 100%
rename from e2e/fixtures/agent-evaluate/evals/automatic-approval.eval.ts
rename to e2e/fixtures/agent-decide/evals/automatic-approval.eval.ts
diff --git a/e2e/fixtures/agent-evaluate/evals/evals.config.ts b/e2e/fixtures/agent-decide/evals/evals.config.ts
similarity index 100%
rename from e2e/fixtures/agent-evaluate/evals/evals.config.ts
rename to e2e/fixtures/agent-decide/evals/evals.config.ts
diff --git a/e2e/fixtures/agent-evaluate/evals/failure.eval.ts b/e2e/fixtures/agent-decide/evals/failure.eval.ts
similarity index 69%
rename from e2e/fixtures/agent-evaluate/evals/failure.eval.ts
rename to e2e/fixtures/agent-decide/evals/failure.eval.ts
index cde789d36..b4d28c223 100644
--- a/e2e/fixtures/agent-evaluate/evals/failure.eval.ts
+++ b/e2e/fixtures/agent-decide/evals/failure.eval.ts
@@ -1,14 +1,14 @@
import { defineEval } from "eve/evals";
export default defineEval({
- description: "An unavailable evaluation model fails routing before any LLM or tool call.",
+ description: "An unavailable decision model fails routing before any LLM or tool call.",
async test(t) {
const turn = await t.send("Alice records the service unavailable incident for review.");
turn.eventsSatisfy("routing failure is reported", (events) =>
events.some(
(event) =>
event.type === "turn.failed" &&
- event.data.message.includes("Evaluation service unavailable"),
+ event.data.message.includes("Decision service unavailable"),
),
);
turn.usedNoTools();
diff --git a/e2e/fixtures/agent-evaluate/evals/judge.eval.ts b/e2e/fixtures/agent-decide/evals/judge.eval.ts
similarity index 91%
rename from e2e/fixtures/agent-evaluate/evals/judge.eval.ts
rename to e2e/fixtures/agent-decide/evals/judge.eval.ts
index 9207d1acf..0ba204e8a 100644
--- a/e2e/fixtures/agent-evaluate/evals/judge.eval.ts
+++ b/e2e/fixtures/agent-decide/evals/judge.eval.ts
@@ -1,12 +1,12 @@
-import type { Experimental_EvaluationModel } from "ai";
+import type { Experimental_DecisionModel } from "ai";
import { defineEval } from "eve/evals";
-const model: Exclude<Experimental_EvaluationModel, string> = {
+const model: Exclude<Experimental_DecisionModel, string> = {
specificationVersion: "v4",
provider: "fixture",
modelId: "fixture-judge",
supportedQuestionTypes: ["boolean", "score", "choice"],
- async doEvaluate({ state, questions }) {
+ async doDecide({ state, questions }) {
const routine = JSON.stringify(state).includes("openai/small");
return {
answers: Object.fromEntries(
@@ -24,7 +24,7 @@ const model: Exclude<Experimental_EvaluationModel, string> = {
};
export default defineEval({
- description: "Single and batched judges score a fixture response through an evaluation model.",
+ description: "Single and batched judges score a fixture response through a decision model.",
judge: { model },
async test(t) {
const turn = await t.send("Alice needs a routine summary of the selected model.");
diff --git a/e2e/fixtures/agent-evaluate/evals/parallel-children.eval.ts b/e2e/fixtures/agent-decide/evals/parallel-children.eval.ts
similarity index 100%
rename from e2e/fixtures/agent-evaluate/evals/parallel-children.eval.ts
rename to e2e/fixtures/agent-decide/evals/parallel-children.eval.ts
diff --git a/e2e/fixtures/agent-evaluate/evals/routing.eval.ts b/e2e/fixtures/agent-decide/evals/routing.eval.ts
similarity index 89%
rename from e2e/fixtures/agent-evaluate/evals/routing.eval.ts
rename to e2e/fixtures/agent-decide/evals/routing.eval.ts
index 97ccf3577..beecd1fc1 100644
--- a/e2e/fixtures/agent-evaluate/evals/routing.eval.ts
+++ b/e2e/fixtures/agent-decide/evals/routing.eval.ts
@@ -1,7 +1,7 @@
import { defineEval } from "eve/evals";
export default defineEval({
- description: "An evaluation model routes each turn once without model tool calls.",
+ description: "A decision model routes each turn once without model tool calls.",
async test(t) {
const first = await t.send("Alice needs a routine summary of the incident evidence.");
first.expectOk();
diff --git a/e2e/fixtures/agent-decide/evals/tool-decision-failure.eval.ts b/e2e/fixtures/agent-decide/evals/tool-decision-failure.eval.ts
new file mode 100644
index 000000000..82b9740af
--- /dev/null
+++ b/e2e/fixtures/agent-decide/evals/tool-decision-failure.eval.ts
@@ -0,0 +1,14 @@
+import { defineEval } from "eve/evals";
+
+export default defineEval({
+ description: "An invalid decision response is reported as a tool error.",
+ async test(t) {
+ const turn = await t.send(
+ "Bob uses decide-request to review a missing answer from the decision service.",
+ );
+ turn.expectOk();
+ turn.calledTool("decide-request", { status: "failed", count: 1 });
+ turn.messageIncludes('"isError":true');
+ t.succeeded();
+ },
+});
diff --git a/e2e/fixtures/agent-evaluate/evals/tool-evaluation.eval.ts b/e2e/fixtures/agent-decide/evals/tool-decision.eval.ts
similarity index 60%
rename from e2e/fixtures/agent-evaluate/evals/tool-evaluation.eval.ts
rename to e2e/fixtures/agent-decide/evals/tool-decision.eval.ts
index 210c66ffa..b1137fc26 100644
--- a/e2e/fixtures/agent-evaluate/evals/tool-evaluation.eval.ts
+++ b/e2e/fixtures/agent-decide/evals/tool-decision.eval.ts
@@ -1,13 +1,13 @@
import { defineEval } from "eve/evals";
export default defineEval({
- description: "A tool evaluates structured state and returns typed answers and usage.",
+ description: "A tool decides structured state and returns typed answers and usage.",
async test(t) {
const turn = await t.send(
- "Alice uses evaluate-request to choose a model for her routine summary.",
+ "Alice uses decide-request to choose a model for her routine summary.",
);
turn.expectOk();
- turn.calledTool("evaluate-request");
+ turn.calledTool("decide-request");
turn.messageIncludes('"isError":false');
turn.messageIncludes('"choice":"openai/small"');
turn.messageIncludes('"totalTokens":45');
diff --git a/e2e/fixtures/agent-evaluate/package.json b/e2e/fixtures/agent-decide/package.json
similarity index 94%
rename from e2e/fixtures/agent-evaluate/package.json
rename to e2e/fixtures/agent-decide/package.json
index da7d59261..32dd919cc 100644
--- a/e2e/fixtures/agent-evaluate/package.json
+++ b/e2e/fixtures/agent-decide/package.json
@@ -1,5 +1,5 @@
{
- "name": "agent-evaluate",
+ "name": "agent-decide",
"version": "0.0.0",
"private": true,
"type": "module",
diff --git a/e2e/fixtures/agent-evaluate/tsconfig.json b/e2e/fixtures/agent-decide/tsconfig.json
similarity index 100%
rename from e2e/fixtures/agent-evaluate/tsconfig.json
rename to e2e/fixtures/agent-decide/tsconfig.json
diff --git a/e2e/fixtures/agent-evaluate/evals/tool-evaluation-failure.eval.ts b/e2e/fixtures/agent-evaluate/evals/tool-evaluation-failure.eval.ts
deleted file mode 100644
index ee08d77b7..000000000
--- a/e2e/fixtures/agent-evaluate/evals/tool-evaluation-failure.eval.ts
+++ /dev/null
@@ -1,14 +0,0 @@
-import { defineEval } from "eve/evals";
-
-export default defineEval({
- description: "An invalid evaluation response is reported as a tool error.",
- async test(t) {
- const turn = await t.send(
- "Bob uses evaluate-request to review a missing answer from the evaluation service.",
- );
- turn.expectOk();
- turn.calledTool("evaluate-request", { status: "failed", count: 1 });
- turn.messageIncludes('"isError":true');
- t.succeeded();
- },
-});
diff --git a/e2e/fixtures/e2e-config/src/judge.ts b/e2e/fixtures/e2e-config/src/judge.ts
index 36ee89b77..3591e00db 100644
--- a/e2e/fixtures/e2e-config/src/judge.ts
+++ b/e2e/fixtures/e2e-config/src/judge.ts
@@ -1,10 +1,11 @@
import { createOpenAI } from "@ai-sdk/openai";
-import type { EveEvalJudgeConfig } from "eve/evals";
+
+type JudgeModel = ReturnType<ReturnType<typeof createOpenAI>["decisionModel"]>;
/** Shared fixture judge, independent of the agent matrix, until CI has access to Jev. */
-export function e2eJudgeModel(): Exclude<EveEvalJudgeConfig["model"], string | undefined> {
+export function e2eJudgeModel(): JudgeModel {
return createOpenAI({
apiKey: process.env.AI_GATEWAY_API_KEY,
baseURL: "https://ai-gateway.vercel.sh/v1",
- }).evaluationModel("openai/gpt-5.6-luna");
+ }).decisionModel("openai/gpt-5.6-luna");
}
diff --git a/e2e/fixtures/e2e-config/test/judge.integration.test.mjs b/e2e/fixtures/e2e-config/test/judge.integration.test.mjs
index 7566d90f4..4b7f427d4 100644
--- a/e2e/fixtures/e2e-config/test/judge.integration.test.mjs
+++ b/e2e/fixtures/e2e-config/test/judge.integration.test.mjs
@@ -3,7 +3,7 @@ import { test } from "node:test";
import { e2eJudgeModel } from "../src/judge.ts";
-test("fixture judge adapts Gateway Responses into evaluation answers", async (t) => {
+test("fixture judge adapts Gateway Responses into decision answers", async (t) => {
const previousKey = process.env.AI_GATEWAY_API_KEY;
process.env.AI_GATEWAY_API_KEY = "fixture-gateway-key";
t.after(() => {
@@ -33,7 +33,7 @@ test("fixture judge adapts Gateway Responses into evaluation answers", async (t)
});
});
- const result = await e2eJudgeModel().doEvaluate({
+ const result = await e2eJudgeModel().doDecide({
state: { response: "Hello Alice" },
questions: {
accurate: { type: "boolean", instructions: "Greets Alice" },
diff --git a/packages/eve/README.md b/packages/eve/README.md
index 23c72a0c8..59942c3d1 100644
--- a/packages/eve/README.md
+++ b/packages/eve/README.md
@@ -171,7 +171,7 @@ These files ship inside the installed package at `node_modules/eve/docs/`:
- [Getting Started](https://eve.dev/docs/getting-started) — install, scaffold, and run locally
- [Project Layout](https://eve.dev/docs/getting-started#project-layout) — every authored directory in depth
- [`agent.ts`](https://eve.dev/docs/agent-config) — agent config reference
-- [Automatic Model Selection](https://eve.dev/docs/guides/evaluate) — choose an agent model for each request
+- [Automatic Model Selection](https://eve.dev/docs/guides/decide) — choose an agent model for each request
- [TypeScript API Reference](https://eve.dev/docs/reference/typescript-api) — complete `define*` and runtime helper reference
- [Vercel Deployment](https://eve.dev/docs/guides/deployment/overview) — deploy to production
diff --git a/packages/eve/extension-contracts/entrypoints/tool.ts b/packages/eve/extension-contracts/entrypoints/tool.ts
index a7a6c7b3e..913eb66bb 100644
--- a/packages/eve/extension-contracts/entrypoints/tool.ts
+++ b/packages/eve/extension-contracts/entrypoints/tool.ts
@@ -26,4 +26,4 @@ export {
type WorkflowToolInput,
type WorkflowToolOptions,
} from "../../src/public/tools/workflow.ts";
-export { evaluate } from "../../src/public/ai/index.ts";
+export { decide } from "../../src/public/ai/index.ts";
diff --git a/packages/eve/extension-contracts/reports/dynamicTool/v72.json b/packages/eve/extension-contracts/reports/dynamicTool/v72.json
new file mode 100644
index 000000000..1255f14eb
--- /dev/null
+++ b/packages/eve/extension-contracts/reports/dynamicTool/v72.json
@@ -0,0 +1,16 @@
+{
+ "kind": "eve-extension-capability-contract",
+ "capability": "dynamicTool",
+ "epoch": 72,
+ "sha256": "ee6951bc35ab704f840869f913ce002390d37ac6b16ee1b8fc5f470f2f0efed5",
+ "exports": [
+ "DynamicToolEntry",
+ "DynamicToolEvents",
+ "DynamicToolResult",
+ "DynamicToolSet",
+ "auto",
+ "defineDurableCallback",
+ "defineDurableSchema",
+ "defineDynamic"
+ ]
+}
diff --git a/packages/eve/extension-contracts/reports/tool/v76.json b/packages/eve/extension-contracts/reports/tool/v76.json
new file mode 100644
index 000000000..11ee9d65a
--- /dev/null
+++ b/packages/eve/extension-contracts/reports/tool/v76.json
@@ -0,0 +1,29 @@
+{
+ "kind": "eve-extension-capability-contract",
+ "capability": "tool",
+ "epoch": 76,
+ "sha256": "76a0f1b120dd6a33bf67aa8208a48e6fbe648cdfceb1e0d0ab34cb4e67207430",
+ "exports": [
+ "AgentRouterInput",
+ "AgentRouterTool",
+ "WorkflowStepToolContext",
+ "WorkflowTool",
+ "WorkflowToolDefinition",
+ "WorkflowToolInput",
+ "WorkflowToolOptions",
+ "agentRouter",
+ "decide",
+ "defaultWebSearch",
+ "defineTool",
+ "defineWorkflowTool",
+ "disableTool",
+ "isDisabledToolSentinel",
+ "isWebSearchToolDefinition",
+ "serializeModelInputSchema",
+ "toolOutput",
+ "toolOutputPart",
+ "toolResultFrom",
+ "webSearch",
+ "workflow"
+ ]
+}
diff --git a/packages/eve/src/ai/evaluate.integration.test.ts b/packages/eve/src/ai/decide.integration.test.ts
similarity index 56%
rename from packages/eve/src/ai/evaluate.integration.test.ts
rename to packages/eve/src/ai/decide.integration.test.ts
index 445d59c2e..1cd374158 100644
--- a/packages/eve/src/ai/evaluate.integration.test.ts
+++ b/packages/eve/src/ai/decide.integration.test.ts
@@ -1,12 +1,12 @@
-import { Experimental_EvaluationMockModelV4 } from "ai/test";
+import { Experimental_DecisionMockModelV4 } from "ai/test";
import { afterEach, beforeEach, describe, expect, expectTypeOf, it, vi } from "vitest";
-import { evaluate } from "#public/ai/index.js";
+import { decide } from "#public/ai/index.js";
-const localEvaluationModel = vi.hoisted(() => vi.fn());
+const localDecisionModel = vi.hoisted(() => vi.fn());
vi.mock("#internal/model-auth/transport.js", async (importOriginal) => ({
...(await importOriginal<typeof import("#internal/model-auth/transport.js")>()),
- localGatewayEvaluationModel: localEvaluationModel,
+ localGatewayDecisionModel: localDecisionModel,
}));
const questions = {
@@ -20,8 +20,8 @@ const questions = {
} as const;
const state = { request: "Alice needs a copy of her invoice." };
-function evaluationModel() {
- const doEvaluate = vi.fn(async () => ({
+function decisionModel() {
+ const doDecide = vi.fn(async () => ({
answers: {
category: { type: "choice" as const, choice: "billing" },
urgent: { type: "boolean" as const, probability: 0.1 },
@@ -31,23 +31,23 @@ function evaluationModel() {
warnings: [],
providerMetadata: { fixture: { requestId: "request-1" } },
}));
- return { doEvaluate, model: new Experimental_EvaluationMockModelV4({ doEvaluate }) };
+ return { doDecide, model: new Experimental_DecisionMockModelV4({ doDecide }) };
}
beforeEach(() => {
- localEvaluationModel.mockReset();
+ localDecisionModel.mockReset();
vi.stubGlobal("AI_SDK_DEFAULT_PROVIDER", undefined);
});
afterEach(() => vi.unstubAllGlobals());
-describe("evaluate", () => {
+describe("decide", () => {
it("defaults to Jev using the local Gateway connection without a session context", async () => {
- const evaluator = evaluationModel();
- localEvaluationModel.mockReturnValue(evaluator.model);
+ const decider = decisionModel();
+ localDecisionModel.mockReturnValue(decider.model);
- const result = await evaluate({ state, questions });
+ const result = await decide({ state, questions });
- expect(localEvaluationModel).toHaveBeenCalledWith("typesafe-ai/jev");
+ expect(localDecisionModel).toHaveBeenCalledWith("typesafe-ai/jev");
expect(result.answers.category.choice).toBe("billing");
expectTypeOf(result.answers.category.choice).toEqualTypeOf<"billing" | "support">();
expectTypeOf(result.answers.urgent.probability).toEqualTypeOf<number>();
@@ -57,29 +57,29 @@ describe("evaluate", () => {
});
it("resolves an explicit model ID through the local connection", async () => {
- localEvaluationModel.mockReturnValue(evaluationModel().model);
- await evaluate({ model: "typesafe-ai/custom", state, questions });
- expect(localEvaluationModel).toHaveBeenCalledWith("typesafe-ai/custom");
+ localDecisionModel.mockReturnValue(decisionModel().model);
+ await decide({ model: "typesafe-ai/custom", state, questions });
+ expect(localDecisionModel).toHaveBeenCalledWith("typesafe-ai/custom");
});
it("preserves the configured default provider for defaults and aliases", async () => {
- const evaluator = evaluationModel();
- const factory = vi.fn(() => evaluator.model);
- vi.stubGlobal("AI_SDK_DEFAULT_PROVIDER", { evaluationModel: factory });
- localEvaluationModel.mockReturnValue(evaluationModel().model);
+ const decider = decisionModel();
+ const factory = vi.fn(() => decider.model);
+ vi.stubGlobal("AI_SDK_DEFAULT_PROVIDER", { decisionModel: factory });
+ localDecisionModel.mockReturnValue(decisionModel().model);
- await evaluate({ state, questions });
- await evaluate({ model: "internal-evaluator", state, questions });
+ await decide({ state, questions });
+ await decide({ model: "internal-decider", state, questions });
- expect(factory.mock.calls).toEqual([["typesafe-ai/jev"], ["internal-evaluator"]]);
- expect(localEvaluationModel).not.toHaveBeenCalled();
+ expect(factory.mock.calls).toEqual([["typesafe-ai/jev"], ["internal-decider"]]);
+ expect(localDecisionModel).not.toHaveBeenCalled();
});
it("passes explicit model instances and request options through unchanged", async () => {
- const evaluator = evaluationModel();
+ const decider = decisionModel();
const abortSignal = new AbortController().signal;
- await evaluate({
- model: evaluator.model,
+ await decide({
+ model: decider.model,
state,
questions,
abortSignal,
@@ -88,8 +88,8 @@ describe("evaluate", () => {
maxRetries: 0,
});
- expect(localEvaluationModel).not.toHaveBeenCalled();
- expect(evaluator.doEvaluate).toHaveBeenCalledWith(
+ expect(localDecisionModel).not.toHaveBeenCalled();
+ expect(decider.doDecide).toHaveBeenCalledWith(
expect.objectContaining({
state,
questions,
@@ -101,36 +101,36 @@ describe("evaluate", () => {
});
it("preserves input and answer validation", async () => {
- const evaluator = evaluationModel();
- await expect(evaluate({ model: evaluator.model, state, questions: {} })).rejects.toThrow();
- expect(evaluator.doEvaluate).not.toHaveBeenCalled();
+ const decider = decisionModel();
+ await expect(decide({ model: decider.model, state, questions: {} })).rejects.toThrow();
+ expect(decider.doDecide).not.toHaveBeenCalled();
- const invalid = new Experimental_EvaluationMockModelV4({
- doEvaluate: async () => ({ answers: {}, warnings: [] }),
+ const invalid = new Experimental_DecisionMockModelV4({
+ doDecide: async () => ({ answers: {}, warnings: [] }),
});
- await expect(evaluate({ model: invalid, state, questions })).rejects.toThrow(
+ await expect(decide({ model: invalid, state, questions })).rejects.toThrow(
"exactly one answer",
);
});
it("propagates provider errors and aborts before provider I/O", async () => {
- const error = new Error("Evaluation unavailable.");
- const failed = new Experimental_EvaluationMockModelV4({
- doEvaluate: async () => {
+ const error = new Error("Decision unavailable.");
+ const failed = new Experimental_DecisionMockModelV4({
+ doDecide: async () => {
throw error;
},
});
- await expect(evaluate({ model: failed, state, questions, maxRetries: 0 })).rejects.toBe(error);
+ await expect(decide({ model: failed, state, questions, maxRetries: 0 })).rejects.toBe(error);
- const evaluator = evaluationModel();
+ const decider = decisionModel();
await expect(
- evaluate({
- model: evaluator.model,
+ decide({
+ model: decider.model,
state,
questions,
abortSignal: AbortSignal.abort(error),
}),
).rejects.toBe(error);
- expect(evaluator.doEvaluate).not.toHaveBeenCalled();
+ expect(decider.doDecide).not.toHaveBeenCalled();
});
});
diff --git a/packages/eve/src/ai/decide.ts b/packages/eve/src/ai/decide.ts
new file mode 100644
index 000000000..9627631d6
--- /dev/null
+++ b/packages/eve/src/ai/decide.ts
@@ -0,0 +1,28 @@
+import {
+ experimental_decide as decideWithAiSdk,
+ type Experimental_DecisionModel as DecisionModel,
+ type Experimental_DecisionQuestion as DecisionQuestion,
+} from "ai";
+
+import { ensureAiSdkWarningLogger } from "#instrumentation/ai-sdk-warnings.js";
+import { localGatewayDecisionModel } from "#internal/model-auth/transport.js";
+
+export const DEFAULT_DECISION_MODEL = "typesafe-ai/jev";
+
+/** Decide typed questions about the state you pass to it using eve's model authentication. */
+export function decide<const Questions extends Record<string, DecisionQuestion>>({
+ model = DEFAULT_DECISION_MODEL,
+ ...options
+}: Omit<Parameters<typeof decideWithAiSdk<Questions>>[0], "model"> & {
+ /** Decision model instance or ID. Defaults to TypeSafe Jev. */
+ model?: DecisionModel;
+}) {
+ ensureAiSdkWarningLogger();
+ return decideWithAiSdk({
+ ...options,
+ model:
+ typeof model === "string" && Reflect.get(globalThis, "AI_SDK_DEFAULT_PROVIDER") == null
+ ? (localGatewayDecisionModel(model) ?? model)
+ : model,
+ });
+}
diff --git a/packages/eve/src/ai/evaluate.ts b/packages/eve/src/ai/evaluate.ts
deleted file mode 100644
index 41faad09d..000000000
--- a/packages/eve/src/ai/evaluate.ts
+++ /dev/null
@@ -1,28 +0,0 @@
-import {
- experimental_evaluate as evaluateWithAiSdk,
- type Experimental_EvaluationModel as EvaluationModel,
- type Experimental_EvaluationQuestion as EvaluationQuestion,
-} from "ai";
-
-import { ensureAiSdkWarningLogger } from "#instrumentation/ai-sdk-warnings.js";
-import { localGatewayEvaluationModel } from "#internal/model-auth/transport.js";
-
-export const DEFAULT_EVALUATION_MODEL = "typesafe-ai/jev";
-
-/** Evaluate typed questions about the state you pass to it using eve's model authentication. */
-export function evaluate<const Questions extends Record<string, EvaluationQuestion>>({
- model = DEFAULT_EVALUATION_MODEL,
- ...options
-}: Omit<Parameters<typeof evaluateWithAiSdk<Questions>>[0], "model"> & {
- /** Evaluation model instance or ID. Defaults to TypeSafe Jev. */
- model?: EvaluationModel;
-}) {
- ensureAiSdkWarningLogger();
- return evaluateWithAiSdk({
- ...options,
- model:
- typeof model === "string" && Reflect.get(globalThis, "AI_SDK_DEFAULT_PROVIDER") == null
- ? (localGatewayEvaluationModel(model) ?? model)
- : model,
- });
-}
diff --git a/packages/eve/src/ai/package.scenario.test.ts b/packages/eve/src/ai/package.scenario.test.ts
index 149e12984..3c5e263c6 100644
--- a/packages/eve/src/ai/package.scenario.test.ts
+++ b/packages/eve/src/ai/package.scenario.test.ts
@@ -8,31 +8,31 @@ import { useScenarioApp } from "#internal/testing/scenario-app.js";
const scenarioApp = useScenarioApp();
-it("builds an agent and child with Gateway and provider evaluation models from packed eve", async () => {
+it("builds an agent and child with Gateway and provider decision models from packed eve", async () => {
const app = await scenarioApp({
name: "ai-models",
installDependencies: true,
files: {
"agent/instructions.md": "Help Alice review the export incident.",
- "agent/evaluation.ts": `import type { Experimental_EvaluationModel } from "ai";
-export const evaluationModel = {
+ "agent/decision.ts": `import type { Experimental_DecisionModel } from "ai";
+export const decisionModel = {
specificationVersion: "v4",
provider: "fixture",
- modelId: "fixture-evaluator",
+ modelId: "fixture-decider",
supportedQuestionTypes: ["choice"],
- async doEvaluate() { throw new Error("Build must not evaluate models."); },
-} satisfies Exclude<Experimental_EvaluationModel, string>;`,
+ async doDecide() { throw new Error("Build must not decide models."); },
+} satisfies Exclude<Experimental_DecisionModel, string>;`,
"agent/agent.ts": `import { defineAgent } from "eve";
import { auto } from "eve/models";
import { anthropic } from "eve/models/anthropic";
export default defineAgent({ model: auto({ options: { "openai/gpt-5.6-sol": "Investigations", my_secret_model: { model: anthropic("sonnet-5"), reasoning: "low", description: "Routine work" } } }) });`,
"agent/tools/classify.ts": `import { defineTool } from "eve/tools";
-import { evaluate } from "eve/ai";
+import { decide } from "eve/ai";
export default defineTool({
description: "Classify an incident",
inputSchema: { type: "object", properties: {}, additionalProperties: false },
async execute(_input, ctx) {
- const result = await evaluate({
+ const result = await decide({
state: { incident: "Alice needs an export summary." },
questions: { category: { type: "choice", instructions: "Choose a category.", criteria: { routine: "Routine work", investigation: "Investigation" } } },
abortSignal: ctx.abortSignal,
@@ -44,8 +44,8 @@ export default defineTool({
"agent/subagents/worker/agent.ts": `import { defineAgent } from "eve";
import { auto } from "eve/models";
import { anthropic } from "eve/models/anthropic";
-import { evaluationModel } from "../../evaluation";
-export default defineAgent({ description: "Review evidence", model: auto({ model: evaluationModel, options: { reviewer: { model: anthropic("sonnet-5"), reasoning: "low", description: "Investigations" } } }) });`,
+import { decisionModel } from "../../decision";
+export default defineAgent({ description: "Review evidence", model: auto({ model: decisionModel, options: { reviewer: { model: anthropic("sonnet-5"), reasoning: "low", description: "Investigations" } } }) });`,
},
});
diff --git a/packages/eve/src/compiler/extension-compatibility.ts b/packages/eve/src/compiler/extension-compatibility.ts
index 1f9ae75bf..1e63f8908 100644
--- a/packages/eve/src/compiler/extension-compatibility.ts
+++ b/packages/eve/src/compiler/extension-compatibility.ts
@@ -22,11 +22,8 @@ interface ExtensionCapabilityContract {
const EXTENSION_CAPABILITY_CONTRACTS = {
extension: { current: 1, supported: [1], dropped: {} },
tool: {
- current: 75,
- supported: [
- 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 29, 30, 31, 32, 34, 35, 54, 55, 63, 64, 65, 66, 67,
- 68, 69, 70, 71, 72, 73, 74, 75,
- ],
+ current: 76,
+ supported: [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 29, 30, 31, 32, 34, 35, 76],
dropped: {
14: "TaskExec.delegated was removed; migrate to workflow-backed background tools",
15: "TaskExec replaces stageEffect with send",
@@ -55,13 +52,15 @@ const EXTENSION_CAPABILITY_CONTRACTS = {
44: "Background defineTool and TaskExec were removed; use defineWorkflowTool for durable background work.",
45: "Background defineTool and TaskExec were removed; use defineWorkflowTool for durable background work.",
46: "Background defineTool and TaskExec were removed; use defineWorkflowTool for durable background work.",
- 47: "eve/experimental/evaluate was removed; import evaluate from eve/ai",
- 48: "eve/experimental/evaluate was removed; import evaluate from eve/ai",
+ 47: "eve/experimental/evaluate was removed; import decide from eve/ai",
+ 48: "eve/experimental/evaluate was removed; import decide from eve/ai",
49: "Background defineTool and TaskExec were removed; use defineWorkflowTool for durable background work.",
50: "Background defineTool and TaskExec were removed; use defineWorkflowTool for durable background work.",
51: "Background defineTool and TaskExec were removed; use defineWorkflowTool for durable background work.",
52: "Background defineTool and TaskExec were removed; use defineWorkflowTool for durable background work.",
53: "Sandbox sessions no longer expose core identity and mutable networking is provider-specific.",
+ 54: "eve/ai evaluate was renamed to decide and AI SDK evaluation models became decision models; import decide from eve/ai",
+ 55: "eve/ai evaluate was renamed to decide and AI SDK evaluation models became decision models; import decide from eve/ai",
56: "SandboxSession no longer exposes setNetworkPolicy; pass the configured environment to ctx.getSandbox(environment) to access network policy capabilities",
57: "agentRouter() no longer accepts a model-facing outputSchema; structured subagent output is requested only through ctx.agent().",
58: "Workflow tools no longer accept execution or return background task receipts; every call blocks until it settles.",
@@ -69,13 +68,26 @@ const EXTENSION_CAPABILITY_CONTRACTS = {
60: "Workflow tools define exactly one of execute(input, ctx), task(input, ctx), or serve(receive, ctx): a task() call returns a receipt and runs as a task, a serve() call reaches a resumable task through receive() and its model input gains taskId, and stream events gained task.started and task.settled. Agent tools are serve() tools with taskId in place of agentId, and the subagent.* stream events were removed. ctx.interruptSignal and WorkflowTaskContext were removed: a steering message aborts the abortSignal of an execute call the turn waits on.",
61: "Approval response policies now run for Cancel as well as Approve; a policy written for Approve only could reject every Cancel.",
62: "Approval response policies now run for Cancel as well as Approve; a policy written for Approve only could reject every Cancel.",
+ 63: "eve/ai evaluate was renamed to decide and AI SDK evaluation models became decision models; import decide from eve/ai",
+ 64: "eve/ai evaluate was renamed to decide and AI SDK evaluation models became decision models; import decide from eve/ai",
+ 65: "eve/ai evaluate was renamed to decide and AI SDK evaluation models became decision models; import decide from eve/ai",
+ 66: "eve/ai evaluate was renamed to decide and AI SDK evaluation models became decision models; import decide from eve/ai",
+ 67: "eve/ai evaluate was renamed to decide and AI SDK evaluation models became decision models; import decide from eve/ai",
+ 68: "eve/ai evaluate was renamed to decide and AI SDK evaluation models became decision models; import decide from eve/ai",
+ 69: "eve/ai evaluate was renamed to decide and AI SDK evaluation models became decision models; import decide from eve/ai",
+ 70: "eve/ai evaluate was renamed to decide and AI SDK evaluation models became decision models; import decide from eve/ai",
+ 71: "eve/ai evaluate was renamed to decide and AI SDK evaluation models became decision models; import decide from eve/ai",
+ 72: "eve/ai evaluate was renamed to decide and AI SDK evaluation models became decision models; import decide from eve/ai",
+ 73: "eve/ai evaluate was renamed to decide and AI SDK evaluation models became decision models; import decide from eve/ai",
+ 74: "eve/ai evaluate was renamed to decide and AI SDK evaluation models became decision models; import decide from eve/ai",
+ 75: "eve/ai evaluate was renamed to decide and AI SDK evaluation models became decision models; import decide from eve/ai",
},
},
dynamicTool: {
- current: 71,
+ current: 72,
supported: [
1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 22, 31, 32, 33, 52, 54,
- 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71,
+ 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 72,
],
dropped: {
21: "Message and reasoning append events now expose deltas instead of cumulative snapshots.",
@@ -111,6 +123,7 @@ const EXTENSION_CAPABILITY_CONTRACTS = {
57: "Stream events gained task.started and task.settled, and input.requested, authorization events, and agent.started carry the taskId of the task they come from. The subagent.called, subagent.completed, subagent.started, and subagent.event stream events were removed; agent calls report task.started, task.settled, and agent.started.",
58: "Approval response policies now run for Cancel as well as Approve; a policy written for Approve only could reject every Cancel.",
59: "Approval response policies now run for Cancel as well as Approve; a policy written for Approve only could reject every Cancel.",
+ 71: "AI SDK evaluation models became decision models; auto() accepts decision models only",
},
},
channel: {
diff --git a/packages/eve/src/evals/define-eval-config.test.ts b/packages/eve/src/evals/define-eval-config.test.ts
index 97f898d73..1d1572ebe 100644
--- a/packages/eve/src/evals/define-eval-config.test.ts
+++ b/packages/eve/src/evals/define-eval-config.test.ts
@@ -82,7 +82,7 @@ describe("defineEvalConfig", () => {
expect(config.judge).toBeUndefined();
});
- it("allows judge settings to use the default evaluation model", () => {
+ it("allows judge settings to use the default decision model", () => {
expect(defineEvalConfig({ judge: {} }).judge).toEqual({});
});
diff --git a/packages/eve/src/evals/define-eval.ts b/packages/eve/src/evals/define-eval.ts
index aa48f2747..c471621f6 100644
--- a/packages/eve/src/evals/define-eval.ts
+++ b/packages/eve/src/evals/define-eval.ts
@@ -17,7 +17,7 @@ import type {
* infer the setup context available as `t.context`.
*
* A `judge` is optional: `t.judge(...)` assertions fall back to the `judge`
- * declared in `evals.config.ts`, then the shared evaluation default. The judge model
+ * declared in `evals.config.ts`, then the shared decision default. The judge model
* is used solely for scoring, never for the agent under test. Eval identity is
* derived from the `evals/<path>.eval.ts` file path by the discovery layer, so
* authoring `id` or `name` throws.
diff --git a/packages/eve/src/evals/judge.integration.test.ts b/packages/eve/src/evals/judge.integration.test.ts
index 0ee4cbac5..a6e3552f7 100644
--- a/packages/eve/src/evals/judge.integration.test.ts
+++ b/packages/eve/src/evals/judge.integration.test.ts
@@ -1,4 +1,4 @@
-import { Experimental_EvaluationMockModelV4 } from "ai/test";
+import { Experimental_DecisionMockModelV4 } from "ai/test";
import { afterEach, describe, expect, it, vi } from "vitest";
import { Client } from "#client/client.js";
import { executeTask } from "#evals/runner/execute-task.js";
@@ -22,8 +22,8 @@ function run(test: (t: EveEvalContext) => void, judge?: EveEvalJudgeConfig, time
});
}
function model(probability = 0.8) {
- const evaluator = new Experimental_EvaluationMockModelV4({
- doEvaluate: async ({ questions }) => ({
+ const decider = new Experimental_DecisionMockModelV4({
+ doDecide: async ({ questions }) => ({
answers: Object.fromEntries(
Object.keys(questions).map((key) => [key, { type: "boolean" as const, probability }]),
),
@@ -31,19 +31,19 @@ function model(probability = 0.8) {
usage: { inputTokens: 10, outputTokens: 2 },
}),
});
- vi.spyOn(evaluator, "doEvaluate");
- return evaluator;
+ vi.spyOn(decider, "doDecide");
+ return decider;
}
afterEach(() => {
vi.unstubAllGlobals();
vi.restoreAllMocks();
});
-describe("evaluation-backed judge runner", () => {
- it("uses evaluate's default model and leaves deterministic evals model-free", async () => {
- const evaluator = model();
- const factory = vi.fn(() => evaluator);
- vi.stubGlobal("AI_SDK_DEFAULT_PROVIDER", { evaluationModel: factory });
+describe("decision-backed judge runner", () => {
+ it("uses decide's default model and leaves deterministic evals model-free", async () => {
+ const decider = model();
+ const factory = vi.fn(() => decider);
+ vi.stubGlobal("AI_SDK_DEFAULT_PROVIDER", { decisionModel: factory });
await run(() => {});
expect(factory).not.toHaveBeenCalled();
const outcome = await run((t) => {
@@ -83,12 +83,12 @@ describe("evaluation-backed judge runner", () => {
},
{ model: inherited, modelOptions: { providerOptions: { fixture: { mode: "inherited" } } } },
);
- expect(vi.mocked(inherited.doEvaluate).mock.calls).toHaveLength(1);
- expect(vi.mocked(overridden.doEvaluate).mock.calls).toHaveLength(1);
- expect(vi.mocked(inherited.doEvaluate).mock.calls[0]?.[0].providerOptions).toEqual({
+ expect(vi.mocked(inherited.doDecide).mock.calls).toHaveLength(1);
+ expect(vi.mocked(overridden.doDecide).mock.calls).toHaveLength(1);
+ expect(vi.mocked(inherited.doDecide).mock.calls[0]?.[0].providerOptions).toEqual({
fixture: { mode: "inherited" },
});
- expect(vi.mocked(overridden.doEvaluate).mock.calls[0]?.[0]).toMatchObject({
+ expect(vi.mocked(overridden.doDecide).mock.calls[0]?.[0]).toMatchObject({
state: { response: "Hello Alice" },
providerOptions: { fixture: { mode: "override" } },
});
@@ -101,7 +101,7 @@ describe("evaluation-backed judge runner", () => {
{ criteria: [], batch: true },
{ criteria: ["Only level"], batch: true },
])("rejects an undersized score rubric before provider I/O: %j", async ({ criteria, batch }) => {
- const evaluator = model();
+ const decider = model();
const outcome = await run(
(t) => {
const question = { type: "score" as const, instructions: "Grade clarity.", criteria };
@@ -117,9 +117,9 @@ describe("evaluation-backed judge runner", () => {
t.judge(question).atLeast(0);
}
},
- { model: evaluator },
+ { model: decider },
);
- expect(evaluator.doEvaluate).not.toHaveBeenCalled();
+ expect(decider.doDecide).not.toHaveBeenCalled();
expect(outcome.assertions).toHaveLength(batch ? 2 : 1);
for (const assertion of outcome.assertions) {
expect(assertion).toMatchObject({
@@ -138,12 +138,12 @@ describe("evaluation-backed judge runner", () => {
it.each(["missing", "invalid", "provider", "unsupported"])(
"fails the whole batch for %s answers or capability",
async (failure) => {
- const evaluator = new Experimental_EvaluationMockModelV4({
+ const decider = new Experimental_DecisionMockModelV4({
supportedQuestionTypes: failure === "unsupported" ? ["choice"] : ["boolean"],
- doEvaluate: async (): Promise<
- Awaited<ReturnType<Experimental_EvaluationMockModelV4["doEvaluate"]>>
+ doDecide: async (): Promise<
+ Awaited<ReturnType<Experimental_DecisionMockModelV4["doDecide"]>>
> => {
- if (failure === "provider") throw new Error("Evaluation credentials unavailable");
+ if (failure === "provider") throw new Error("Decision credentials unavailable");
return {
answers:
failure === "missing"
@@ -165,7 +165,7 @@ describe("evaluation-backed judge runner", () => {
},
});
},
- { model: evaluator },
+ { model: decider },
);
expect(outcome.skipReason).toBeUndefined();
expect(outcome.assertions).toHaveLength(2);
@@ -177,18 +177,18 @@ describe("evaluation-backed judge runner", () => {
);
it("bounds finalization when a provider ignores the eval deadline", async () => {
- const evaluator = new Experimental_EvaluationMockModelV4({
- doEvaluate: () => new Promise(() => {}),
+ const decider = new Experimental_DecisionMockModelV4({
+ doDecide: () => new Promise(() => {}),
});
- vi.spyOn(evaluator, "doEvaluate");
+ vi.spyOn(decider, "doDecide");
const outcome = await run(
(t) => {
t.judge("A");
},
- { model: evaluator },
+ { model: decider },
20,
);
- expect(vi.mocked(evaluator.doEvaluate).mock.calls[0]?.[0].abortSignal?.aborted).toBe(true);
+ expect(vi.mocked(decider.doDecide).mock.calls[0]?.[0].abortSignal?.aborted).toBe(true);
expect(outcome.assertions[0]).toMatchObject({ passed: false, severity: "gate" });
expect(computeEvalVerdict(outcome)).toBe("failed");
});
diff --git a/packages/eve/src/evals/judge.test.ts b/packages/eve/src/evals/judge.test.ts
index 920c3674c..89907e76b 100644
--- a/packages/eve/src/evals/judge.test.ts
+++ b/packages/eve/src/evals/judge.test.ts
@@ -4,8 +4,8 @@ import { buildJudgeContext } from "#evals/judge.js";
import { createEmptyDerivedFacts } from "#evals/runner/derive-run-facts.js";
import type { AssertionHandle, EveEvalTaskResult, JudgeContext } from "#evals/types.js";
-const evaluate = vi.hoisted(() => vi.fn());
-vi.mock("#ai/evaluate.js", () => ({ evaluate }));
+const decide = vi.hoisted(() => vi.fn());
+vi.mock("#ai/decide.js", () => ({ decide }));
afterEach(() => vi.resetAllMocks());
function setup() {
@@ -23,7 +23,7 @@ function setup() {
function result(answers: Record<string, unknown>) {
return {
answers,
- response: { modelId: "test-evaluator", timestamp: new Date(0) },
+ response: { modelId: "test-decider", timestamp: new Date(0) },
usage: { inputTokens: 10, outputTokens: 2, totalTokens: 12 },
warnings: [],
};
@@ -41,11 +41,11 @@ function emptyTaskResult(): EveEvalTaskResult {
describe("judge", () => {
it("expands criteria and retains assertion handles and diagnostics", async () => {
- evaluate.mockResolvedValue(result({ judgment: { type: "boolean", probability: 0.7 } }));
+ decide.mockResolvedValue(result({ judgment: { type: "boolean", probability: 0.7 } }));
const { judge, collector, controller } = setup();
const handle = judge("cites a source").label("citation").atLeast(0.8);
expect(handle).not.toHaveProperty("then");
- expect(evaluate).toHaveBeenCalledWith(
+ expect(decide).toHaveBeenCalledWith(
expect.objectContaining({
model: undefined,
abortSignal: controller.signal,
@@ -60,13 +60,13 @@ describe("judge", () => {
severity: "soft",
threshold: 0.8,
passed: false,
- metadata: { judge: "test-evaluator", answer: { probability: 0.7 }, usageScope: "assertion" },
+ metadata: { judge: "test-decider", answer: { probability: 0.7 }, usageScope: "assertion" },
});
expect(assertion?.message).toContain('"cites a source"');
});
it("shares a batch request, normalizes rubrics, and keeps expectations local", async () => {
- evaluate.mockResolvedValue(
+ decide.mockResolvedValue(
result({
accurate: { type: "boolean", probability: 0.2 },
clarity: { type: "score", score: 1.5 },
@@ -93,8 +93,8 @@ describe("judge", () => {
});
handles.clarity.atLeast(0.8);
handles.outcome.gate();
- expect(evaluate).toHaveBeenCalledOnce();
- expect(evaluate.mock.calls[0]?.[0].questions.outcome).not.toHaveProperty("expected");
+ expect(decide).toHaveBeenCalledOnce();
+ expect(decide.mock.calls[0]?.[0].questions.outcome).not.toHaveProperty("expected");
const assertions = await collector.finalize(emptyTaskResult());
expect(assertions.map(({ name, score, passed }) => ({ name, score, passed }))).toEqual([
{ name: "judge.boolean.accurate", score: 0.2, passed: true },
@@ -105,7 +105,7 @@ describe("judge", () => {
});
it("scores a different selected choice as zero", async () => {
- evaluate.mockResolvedValue(result({ judgment: { type: "choice", choice: "declined" } }));
+ decide.mockResolvedValue(result({ judgment: { type: "choice", choice: "declined" } }));
const { judge, collector } = setup();
judge({
type: "choice",
@@ -120,7 +120,7 @@ describe("judge", () => {
});
it("captures JSON state and questions before later mutations", async () => {
- evaluate.mockResolvedValue(result({ judgment: { type: "boolean", probability: 1 } }));
+ decide.mockResolvedValue(result({ judgment: { type: "boolean", probability: 1 } }));
const { judge, collector } = setup();
const on = { values: [1] };
const question = { type: "boolean" as const, instructions: "Original" };
@@ -128,19 +128,19 @@ describe("judge", () => {
on.values.push(2);
question.instructions = "Changed";
await collector.finalize(emptyTaskResult());
- expect(evaluate.mock.calls[0]?.[0]).toMatchObject({
+ expect(decide.mock.calls[0]?.[0]).toMatchObject({
state: { output: { values: [1] } },
questions: { judgment: { instructions: "Original" } },
});
});
it("preserves explicit null and transcript values", async () => {
- evaluate.mockResolvedValue(result({ judgment: { type: "boolean", probability: 1 } }));
+ decide.mockResolvedValue(result({ judgment: { type: "boolean", probability: 1 } }));
const { judge, collector } = setup();
judge("check", { on: null });
judge("check", { on: "user: Hello\nassistant: Hi" });
await collector.finalize(emptyTaskResult());
- expect(evaluate.mock.calls.map(([request]) => request.state.output)).toEqual([
+ expect(decide.mock.calls.map(([request]) => request.state.output)).toEqual([
null,
"user: Hello\nassistant: Hi",
]);
@@ -154,11 +154,11 @@ describe("judge", () => {
severity: "gate",
passed: false,
});
- expect(evaluate).not.toHaveBeenCalled();
+ expect(decide).not.toHaveBeenCalled();
});
it("turns shared request errors into failed gates for every question", async () => {
- evaluate.mockRejectedValue(new Error("Evaluation unavailable"));
+ decide.mockRejectedValue(new Error("Decision unavailable"));
const { judge, collector } = setup();
const handles = judge({
questions: {
@@ -172,18 +172,18 @@ describe("judge", () => {
expect.objectContaining({
severity: "gate",
passed: false,
- message: "Evaluation unavailable",
+ message: "Decision unavailable",
}),
expect.objectContaining({
severity: "gate",
passed: false,
- message: "Evaluation unavailable",
+ message: "Decision unavailable",
}),
]);
});
it("does not hang when a provider ignores cancellation", async () => {
- evaluate.mockReturnValue(new Promise(() => {}));
+ decide.mockReturnValue(new Promise(() => {}));
const { judge, collector, controller } = setup();
judge("check");
controller.abort(new Error("Deadline reached"));
@@ -196,7 +196,7 @@ describe("judge", () => {
it("rejects empty batches", () => {
expect(() => setup().judge({ questions: {} })).toThrow("non-empty");
- expect(evaluate).not.toHaveBeenCalled();
+ expect(decide).not.toHaveBeenCalled();
});
});
diff --git a/packages/eve/src/evals/judge.ts b/packages/eve/src/evals/judge.ts
index 432dad8e4..ae0d67b64 100644
--- a/packages/eve/src/evals/judge.ts
+++ b/packages/eve/src/evals/judge.ts
@@ -1,4 +1,4 @@
-import { evaluate } from "#ai/evaluate.js";
+import { decide } from "#ai/decide.js";
import { runUntilAborted } from "#evals/abort.js";
import type { AssertionCollector } from "#evals/assertions/collector.js";
import { formatDiagnosticValue, toDiagnosticMetadataValue } from "#evals/diagnostics.js";
@@ -21,7 +21,7 @@ interface JudgeDeps {
readonly signal: AbortSignal;
}
-/** Bind evaluation-backed judgments to the existing assertion lifecycle. */
+/** Bind decision-backed judgments to the existing assertion lifecycle. */
export function buildJudgeContext(deps: JudgeDeps): JudgeContext {
function judge(criteria: string, opts?: JudgeOpts): AssertionHandle;
function judge<const Q extends JudgeQuestion>(
@@ -58,20 +58,20 @@ export function buildJudgeContext(deps: JudgeDeps): JudgeContext {
(async () => {
const capturedState = structuredClone(state);
const capturedQuestions = structuredClone(questions);
- const evaluationQuestions = Object.fromEntries(
+ const decisionQuestions = Object.fromEntries(
Object.entries(capturedQuestions).map(([id, question]) => {
if (question.type !== "choice") return [id, question];
- const { expected, ...evaluationQuestion } = question;
+ const { expected, ...decisionQuestion } = question;
if (typeof expected !== "string" || !Object.hasOwn(question.criteria, expected)) {
throw new Error(`Judge question "${id}" expected must name an option in criteria.`);
}
- return [id, evaluationQuestion];
+ return [id, decisionQuestion];
}),
);
- const result = await evaluate({
+ const result = await decide({
model,
state: capturedState,
- questions: evaluationQuestions,
+ questions: decisionQuestions,
providerOptions: modelOptions?.providerOptions,
abortSignal: deps.signal,
});
@@ -130,7 +130,7 @@ function formatJudgeDetail(state: JudgeInput, question: JudgeQuestion, answer: u
function judgeScore(
question: JudgeQuestion,
- answer: Awaited<ReturnType<typeof evaluate>>["answers"][string],
+ answer: Awaited<ReturnType<typeof decide>>["answers"][string],
): number {
switch (answer.type) {
case "boolean":
@@ -142,5 +142,5 @@ function judgeScore(
if (question.type === "choice") return answer.choice === question.expected ? 1 : 0;
break;
}
- throw new Error("Evaluation answer does not match the judge question.");
+ throw new Error("Decision answer does not match the judge question.");
}
diff --git a/packages/eve/src/evals/session.test.ts b/packages/eve/src/evals/session.test.ts
index 0b5642096..eaa9ce233 100644
--- a/packages/eve/src/evals/session.test.ts
+++ b/packages/eve/src/evals/session.test.ts
@@ -10,10 +10,10 @@ import { EvalSessionManager } from "#evals/session-manager.js";
import { createEvalTargetHandle } from "#evals/target.js";
import { stampTestEvents } from "#internal/testing/events.js";
-const mocks = vi.hoisted(() => ({ evaluate: vi.fn() }));
+const mocks = vi.hoisted(() => ({ decide: vi.fn() }));
vi.mock("node:fs/promises", () => ({ readFile: vi.fn() }));
-vi.mock("#ai/evaluate.js", () => ({ evaluate: mocks.evaluate }));
+vi.mock("#ai/decide.js", () => ({ decide: mocks.decide }));
afterEach(() => {
vi.restoreAllMocks();
@@ -38,7 +38,7 @@ describe("eval judge input", () => {
}
context.judge("answers the user's request");
- expect(mocks.evaluate).toHaveBeenCalledWith(
+ expect(mocks.decide).toHaveBeenCalledWith(
expect.objectContaining({
state: expect.objectContaining({ input: "Find Bob's order status instead." }),
}),
@@ -53,7 +53,7 @@ describe("eval judge input", () => {
await session.sendFile("Summarize Alice's invoice.", "/invoice.txt", "text/plain");
context.judge("summarizes the invoice");
- expect(mocks.evaluate).toHaveBeenCalledWith(
+ expect(mocks.decide).toHaveBeenCalledWith(
expect.objectContaining({
state: expect.objectContaining({ input: "Summarize Alice's invoice." }),
}),
@@ -74,7 +74,7 @@ describe("eval judge input", () => {
{ type: "text", text: "List the line items." },
]);
context.judge("lists the line items");
- expect(mocks.evaluate).toHaveBeenLastCalledWith(
+ expect(mocks.decide).toHaveBeenLastCalledWith(
expect.objectContaining({
state: expect.objectContaining({ input: "Review Alice's invoice.\nList the line items." }),
}),
@@ -82,7 +82,7 @@ describe("eval judge input", () => {
await session.send([file]);
context.judge("describes the attachment");
- expect(mocks.evaluate).toHaveBeenLastCalledWith(
+ expect(mocks.decide).toHaveBeenLastCalledWith(
expect.objectContaining({ state: expect.objectContaining({ input: "" }) }),
);
});
@@ -119,7 +119,7 @@ describe("eval session compact", () => {
});
async function setup() {
- mocks.evaluate.mockResolvedValue({
+ mocks.decide.mockResolvedValue({
answers: { judgment: { type: "boolean", probability: 1 } },
response: { modelId: "test" },
});
diff --git a/packages/eve/src/evals/types.ts b/packages/eve/src/evals/types.ts
index 764bc6e09..9cc457e1f 100644
--- a/packages/eve/src/evals/types.ts
+++ b/packages/eve/src/evals/types.ts
@@ -1,5 +1,5 @@
import type { TokenUsage } from "#shared/token-usage.js";
-import type { Experimental_EvaluationModel as EvaluationModel } from "ai";
+import type { Experimental_DecisionModel as DecisionModel } from "ai";
import type { StandardSchemaV1 } from "#compiled/@standard-schema/spec/index.js";
import type { ClientAgentSession } from "#client/agent-session.js";
@@ -392,14 +392,14 @@ export interface EveEvalTurn extends EveEvalAssertions, EveEvalOutputAssertions
// Judge (LLM-as-judge)
// ---------------------------------------------------------------------------
-/** Evaluation settings used only for scoring, independently of the agent under test. */
+/** Decision settings used only for scoring, independently of the agent under test. */
export interface EveEvalJudgeConfig {
- /** Evaluation model ID or instance. Defaults to the model used by `eve/ai` evaluate. */
- readonly model?: EvaluationModel;
+ /** Decision model ID or instance. Defaults to the model used by `eve/ai` decide. */
+ readonly model?: DecisionModel;
readonly modelOptions?: AgentModelOptionsDefinition;
}
-/** JSON content accepted as evaluation state, instructions, or rubric descriptions. */
+/** JSON content accepted as decision state, instructions, or rubric descriptions. */
export type JudgeInput = string | JsonObject | readonly JsonValue[];
/** A boolean judgment, ordered rubric, or categorical judgment with an expected option. */
@@ -426,7 +426,7 @@ export type JudgeQuestionConstraint<Q extends JudgeQuestion> = Q extends { type:
? { readonly expected: NoInfer<Extract<keyof Q["criteria"], string>> }
: unknown;
-/** Named judgments evaluated together against one shared state. */
+/** Named judgments decided together against one shared state. */
export interface JudgeBatch<Questions extends Record<string, JudgeQuestion>> {
/** Replaces the default `{ input, output }` state when supplied. */
readonly state?: JudgeInput;
@@ -557,7 +557,7 @@ interface EveEvalBase {
/**
* Judge model for this eval's `t.judge(...)` assertions. Optional: when
* omitted, judge assertions fall back to the `judge` declared in
- * `evals.config.ts`, then the shared evaluation default. Only used for
+ * `evals.config.ts`, then the shared decision default. Only used for
* scoring; never changes the agent under test.
*/
readonly judge?: EveEvalJudgeConfig;
@@ -684,7 +684,7 @@ export interface EveEvalConfigInput<TContext = unknown> {
teardown?(context: TContext | undefined): void | Promise<void>;
/**
* Default judge model for `t.judge(...)` assertions across every eval.
- * Optional: omission uses the shared evaluation default. Individual evals
+ * Optional: omission uses the shared decision default. Individual evals
* may override it with their own `judge`. Only ever used for scoring.
*/
readonly judge?: EveEvalJudgeConfig;
diff --git a/packages/eve/src/execution/tools/agent-router-workflow.ts b/packages/eve/src/execution/tools/agent-router-workflow.ts
index 993bfbde8..95264aa0b 100644
--- a/packages/eve/src/execution/tools/agent-router-workflow.ts
+++ b/packages/eve/src/execution/tools/agent-router-workflow.ts
@@ -1,4 +1,4 @@
-import { evaluate } from "#ai/evaluate.js";
+import { decide } from "#ai/decide.js";
import type { JsonValue } from "#shared/json.js";
import type { WorkflowToolContext } from "#tools/workflow-definition.js";
import type { AgentRouterInput } from "#execution/tools/agent-router.js";
@@ -32,7 +32,7 @@ async function chooseTarget(
}
if (names.length === 1) return names[0]!;
- const result = await evaluate({
+ const result = await decide({
abortSignal,
state: { message },
questions: {
diff --git a/packages/eve/src/internal/aisdk-provider-contract.test.ts b/packages/eve/src/internal/aisdk-provider-contract.test.ts
index f2eb6f06b..a51f10ac8 100644
--- a/packages/eve/src/internal/aisdk-provider-contract.test.ts
+++ b/packages/eve/src/internal/aisdk-provider-contract.test.ts
@@ -25,7 +25,7 @@ describe("ai-sdk provider identity contract (canary)", () => {
(globalThis as { AI_SDK_DEFAULT_PROVIDER?: unknown }).AI_SDK_DEFAULT_PROVIDER,
).toBeUndefined();
expect(gateway.languageModel("anthropic/claude-sonnet-5").provider).toBe("gateway");
- expect(gateway.evaluationModel("typesafe-ai/jev").provider).toBe("gateway");
+ expect(gateway.decisionModel("typesafe-ai/jev").provider).toBe("gateway");
});
it("direct provider instances do not report 'gateway'", () => {
diff --git a/packages/eve/src/internal/model-auth/development-broker.integration.test.ts b/packages/eve/src/internal/model-auth/development-broker.integration.test.ts
index 8d2f149c0..6fda07093 100644
--- a/packages/eve/src/internal/model-auth/development-broker.integration.test.ts
+++ b/packages/eve/src/internal/model-auth/development-broker.integration.test.ts
@@ -25,7 +25,7 @@ import { handleDevelopmentModelCredentialRequest } from "./development-broker-se
import { readDevelopmentModelCredential } from "./development-broker-client.js";
import {
createDirectModelFetch,
- localGatewayEvaluationModel,
+ localGatewayDecisionModel,
localGatewayModel,
} from "./transport.js";
import { createCodexFetch } from "#public/models/openai/chatgpt/transport.js";
@@ -224,9 +224,9 @@ it("streams through the same Gateway model after switching teams and then to a k
}
});
-it("evaluates through the Gateway connection selected by /login", async () => {
- const model = localGatewayEvaluationModel("typesafe-ai/jev");
- if (!model) throw new Error("Expected a Gateway evaluation model");
+it("decides through the Gateway connection selected by /login", async () => {
+ const model = localGatewayDecisionModel("typesafe-ai/jev");
+ if (!model) throw new Error("Expected a Gateway decision model");
expect(JSON.stringify(model)).not.toContain("account-token");
expect(mocks.gateway).not.toHaveBeenCalled();
upstream.mockResolvedValue(
@@ -236,7 +236,7 @@ it("evaluates through the Gateway connection selected by /login", async () => {
}),
);
- const result = await model.doEvaluate({
+ const result = await model.doDecide({
state: "Alice requests a routine summary.",
questions: {
route: {
@@ -249,7 +249,7 @@ it("evaluates through the Gateway connection selected by /login", async () => {
expect(result.answers.route).toEqual({ type: "choice", choice: "fast" });
const [url, init] = upstream.mock.calls[0]!;
- expect(String(url)).toBe("https://ai-gateway.vercel.sh/v4/ai/evaluation-model");
+ expect(String(url)).toBe("https://ai-gateway.vercel.sh/v4/ai/decision-model");
expect(init.headers.get("authorization")).toBe("Bearer account-token");
expect(init.headers.get("x-vercel-ai-gateway-team")).toBe("team_a");
expect(init.headers.get("ai-model-id")).toBe("typesafe-ai/jev");
diff --git a/packages/eve/src/internal/model-auth/transport.ts b/packages/eve/src/internal/model-auth/transport.ts
index 19597a6ca..ca8c977ec 100644
--- a/packages/eve/src/internal/model-auth/transport.ts
+++ b/packages/eve/src/internal/model-auth/transport.ts
@@ -1,8 +1,4 @@
-import {
- createGateway,
- type Experimental_EvaluationModel as EvaluationModel,
- type LanguageModel,
-} from "ai";
+import { createGateway, type LanguageModel } from "ai";
import { isEveDevEnvironment } from "#internal/application/dev-environment.js";
import { resolveModelApiKey } from "./api-key.js";
import {
@@ -73,9 +69,9 @@ export function localGatewayModel(id: string): LanguageModel | undefined {
return createLocalGateway()?.languageModel(id);
}
-/** Resolve an evaluation model through the same local Gateway connection used by language models. */
-export function localGatewayEvaluationModel(
+/** Resolve a decision model through the same local Gateway connection used by language models. */
+export function localGatewayDecisionModel(
id: string,
-): Exclude<EvaluationModel, string> | undefined {
- return createLocalGateway()?.evaluationModel(id);
+): ReturnType<ReturnType<typeof createGateway>["decisionModel"]> | undefined {
+ return createLocalGateway()?.decisionModel(id);
}
diff --git a/packages/eve/src/internal/nitro/host/build-extension.scenario.test.ts b/packages/eve/src/internal/nitro/host/build-extension.scenario.test.ts
index 26ec553b8..bd7f62b9a 100644
--- a/packages/eve/src/internal/nitro/host/build-extension.scenario.test.ts
+++ b/packages/eve/src/internal/nitro/host/build-extension.scenario.test.ts
@@ -159,25 +159,25 @@ describe("extension build output", () => {
});
it.each([
- ["static re-export", 'export { evaluate as assess } from "eve/ai";'],
+ ["static re-export", 'export { decide as assess } from "eve/ai";'],
[
"namespace import",
- 'import * as evaluation from "eve/ai"; export const assess = evaluation.evaluate;',
+ 'import * as decision from "eve/ai"; export const assess = decision.decide;',
],
[
"dynamic import",
- 'export async function assess(options: Parameters<typeof import("eve/ai").evaluate>[0]): Promise<void> { await (await import("eve/ai")).evaluate(options); }',
+ 'export async function assess(options: Parameters<typeof import("eve/ai").decide>[0]): Promise<void> { await (await import("eve/ai")).decide(options); }',
],
])("stamps the tool capability for a hook-only extension using a %s", async (_name, helper) => {
const root = await createExtensionPackage();
await rm(join(root, "extension", "tools"), { recursive: true });
await mkdir(join(root, "extension", "hooks"));
await mkdir(join(root, "extension", "lib"));
- await writeFile(join(root, "extension", "lib", "evaluation.ts"), helper);
+ await writeFile(join(root, "extension", "lib", "decision.ts"), helper);
await writeFile(
- join(root, "extension", "hooks", "evaluate.ts"),
+ join(root, "extension", "hooks", "decide.ts"),
`import { defineHook } from "eve/hooks";
-import { assess } from "../lib/evaluation";
+import { assess } from "../lib/decision";
export default defineHook({ events: { "turn.started": async () => {
await assess({ state: { request: "Alice needs a summary." }, questions: {
routine: { type: "boolean", instructions: "Is this routine work?" }
@@ -221,7 +221,7 @@ export default defineHook({ events: { "turn.started": async () => {
],
[
"type-only AI import",
- 'import type { evaluate } from "eve/ai"; export type Evaluate = typeof evaluate;',
+ 'import type { decide } from "eve/ai"; export type Decide = typeof decide;',
undefined,
],
[
@@ -235,7 +235,7 @@ export default defineHook({ events: { "turn.started": async () => {
const root = await createExtensionPackage();
await rm(join(root, "extension", "tools"), { recursive: true });
await mkdir(join(root, "extension", "lib"));
- await writeFile(join(root, "extension", "lib", "evaluation.ts"), helper);
+ await writeFile(join(root, "extension", "lib", "decision.ts"), helper);
const config = await tryReadExtensionBuildConfig(root);
const outDir = await buildExtensionPackage(root, config!);
const manifestPath = join(outDir, "extension", "_manifest.json");
diff --git a/packages/eve/src/models/auto.integration.test.ts b/packages/eve/src/models/auto.integration.test.ts
index 4d278b67e..0f29ff5f9 100644
--- a/packages/eve/src/models/auto.integration.test.ts
+++ b/packages/eve/src/models/auto.integration.test.ts
@@ -1,4 +1,4 @@
-import { Experimental_EvaluationMockModelV4 } from "ai/test";
+import { Experimental_DecisionMockModelV4 } from "ai/test";
import { beforeEach, describe, expect, it, vi } from "vitest";
import { ContextContainer } from "#context/container.js";
@@ -9,7 +9,7 @@ import { anthropic } from "#public/models/anthropic/index.js";
import { auto } from "./auto.js";
const runtime = vi.hoisted(() => ({
- localEvaluationModel: vi.fn(),
+ localDecisionModel: vi.fn(),
logWarn: vi.fn(),
state: undefined as ContextContainer | undefined,
}));
@@ -28,7 +28,7 @@ vi.mock("#internal/logging.js", async (importOriginal) => ({
}));
vi.mock("#internal/model-auth/transport.js", async (importOriginal) => ({
...(await importOriginal<typeof import("#internal/model-auth/transport.js")>()),
- localGatewayEvaluationModel: runtime.localEvaluationModel,
+ localGatewayDecisionModel: runtime.localDecisionModel,
}));
const options = {
@@ -53,36 +53,36 @@ function event(turnId = "turn_1") {
return { type: "step.started", data: { turnId } };
}
-function evaluationModel(choice = "openai/small", modelId = "fixture-evaluator") {
- const doEvaluate = vi.fn(async () => ({
+function decisionModel(choice = "openai/small", modelId = "fixture-decider") {
+ const doDecide = vi.fn(async () => ({
answers: { route: { type: "choice" as const, choice } },
usage: { inputTokens: 42, outputTokens: 3 },
warnings: [],
response: { modelId },
}));
return {
- doEvaluate,
- model: new Experimental_EvaluationMockModelV4({ modelId, doEvaluate }),
+ doDecide,
+ model: new Experimental_DecisionMockModelV4({ modelId, doDecide }),
};
}
beforeEach(() => {
runtime.state = new ContextContainer();
- runtime.localEvaluationModel.mockReset();
+ runtime.localDecisionModel.mockReset();
runtime.logWarn.mockReset();
});
describe("auto", () => {
it("defaults to Jev through the AI SDK default provider", async () => {
- const evaluator = evaluationModel();
- const evaluationModelFactory = vi.fn(() => evaluator.model);
+ const decider = decisionModel();
+ const decisionModelFactory = vi.fn(() => decider.model);
const previous = Reflect.get(globalThis, "AI_SDK_DEFAULT_PROVIDER");
- Reflect.set(globalThis, "AI_SDK_DEFAULT_PROVIDER", { evaluationModel: evaluationModelFactory });
+ Reflect.set(globalThis, "AI_SDK_DEFAULT_PROVIDER", { decisionModel: decisionModelFactory });
try {
const handler = auto({ options }).events["step.started"]!;
await expect(handler(event(), context())).resolves.toBe("openai/small");
- expect(runtime.localEvaluationModel).not.toHaveBeenCalled();
- expect(evaluationModelFactory).toHaveBeenCalledWith("typesafe-ai/jev");
+ expect(runtime.localDecisionModel).not.toHaveBeenCalled();
+ expect(decisionModelFactory).toHaveBeenCalledWith("typesafe-ai/jev");
} finally {
if (previous === undefined) Reflect.deleteProperty(globalThis, "AI_SDK_DEFAULT_PROVIDER");
else Reflect.set(globalThis, "AI_SDK_DEFAULT_PROVIDER", previous);
@@ -90,41 +90,41 @@ describe("auto", () => {
});
it("preserves a custom default provider when the local Gateway connection is available", async () => {
- const evaluator = evaluationModel();
- const localEvaluator = evaluationModel("openai/large");
- runtime.localEvaluationModel.mockReturnValue(localEvaluator.model);
- const evaluationModelFactory = vi.fn(() => evaluator.model);
+ const decider = decisionModel();
+ const localDecider = decisionModel("openai/large");
+ runtime.localDecisionModel.mockReturnValue(localDecider.model);
+ const decisionModelFactory = vi.fn(() => decider.model);
const previous = Reflect.get(globalThis, "AI_SDK_DEFAULT_PROVIDER");
- Reflect.set(globalThis, "AI_SDK_DEFAULT_PROVIDER", { evaluationModel: evaluationModelFactory });
+ Reflect.set(globalThis, "AI_SDK_DEFAULT_PROVIDER", { decisionModel: decisionModelFactory });
try {
const handler = auto({ model: "internal-router", options }).events["step.started"]!;
await expect(handler(event(), context())).resolves.toBe("openai/small");
- expect(evaluationModelFactory).toHaveBeenCalledWith("internal-router");
- expect(evaluator.doEvaluate).toHaveBeenCalledOnce();
- expect(runtime.localEvaluationModel).not.toHaveBeenCalled();
- expect(localEvaluator.doEvaluate).not.toHaveBeenCalled();
+ expect(decisionModelFactory).toHaveBeenCalledWith("internal-router");
+ expect(decider.doDecide).toHaveBeenCalledOnce();
+ expect(runtime.localDecisionModel).not.toHaveBeenCalled();
+ expect(localDecider.doDecide).not.toHaveBeenCalled();
} finally {
if (previous === undefined) Reflect.deleteProperty(globalThis, "AI_SDK_DEFAULT_PROVIDER");
else Reflect.set(globalThis, "AI_SDK_DEFAULT_PROVIDER", previous);
}
});
- it("uses the local Gateway connection for a string evaluation model", async () => {
- const evaluator = evaluationModel();
- runtime.localEvaluationModel.mockReturnValue(evaluator.model);
+ it("uses the local Gateway connection for a string decision model", async () => {
+ const decider = decisionModel();
+ runtime.localDecisionModel.mockReturnValue(decider.model);
const handler = auto({ options }).events["step.started"]!;
await expect(handler(event(), context())).resolves.toBe("openai/small");
- expect(runtime.localEvaluationModel).toHaveBeenCalledWith("typesafe-ai/jev");
- expect(evaluator.doEvaluate).toHaveBeenCalledOnce();
+ expect(runtime.localDecisionModel).toHaveBeenCalledWith("typesafe-ai/jev");
+ expect(decider.doDecide).toHaveBeenCalledOnce();
});
it("routes provider language models by alias and preserves reasoning", async () => {
const languageModel = anthropic("sonnet-5");
- const evaluator = evaluationModel("private");
+ const decider = decisionModel("private");
const handler = auto({
- model: evaluator.model,
+ model: decider.model,
options: {
...options,
private: {
@@ -139,7 +139,7 @@ describe("auto", () => {
model: languageModel,
reasoning: "low",
});
- expect(evaluator.doEvaluate).toHaveBeenCalledWith(
+ expect(decider.doDecide).toHaveBeenCalledWith(
expect.objectContaining({
state: { messages: [{ role: "user", text: "Alice requests a routine summary." }] },
questions: {
@@ -154,27 +154,27 @@ describe("auto", () => {
);
});
- it("evaluates once per turn and restores the selection from durable context", async () => {
- const evaluator = evaluationModel();
- const handler = auto({ model: evaluator.model, options }).events["step.started"]!;
+ it("decides once per turn and restores the selection from durable context", async () => {
+ const decider = decisionModel();
+ const handler = auto({ model: decider.model, options }).events["step.started"]!;
await handler(event(), context());
runtime.state = await deserializeContext(serializeContext(runtime.state!));
await handler(event(), context());
await handler(event("turn_2"), context());
- expect(evaluator.doEvaluate).toHaveBeenCalledTimes(2);
+ expect(decider.doDecide).toHaveBeenCalledTimes(2);
expect(Object.keys(serializeContext(runtime.state!))).toEqual([
- expect.stringMatching(/^eve\.experimental\.evaluate\.model\./),
+ expect.stringMatching(/^eve\.experimental\.decide\.model\./),
]);
});
- it("uses and retains the fallback model when evaluation fails", async () => {
- const providerError = new Error("evaluation unavailable");
- const doEvaluate = vi.fn(async () => {
+ it("uses and retains the fallback model when decision fails", async () => {
+ const providerError = new Error("decision unavailable");
+ const doDecide = vi.fn(async () => {
throw providerError;
});
- const failed = new Experimental_EvaluationMockModelV4({ doEvaluate });
+ const failed = new Experimental_DecisionMockModelV4({ doDecide });
const handler = auto({
model: failed,
fallback: "anthropic/claude-sonnet-5",
@@ -183,11 +183,11 @@ describe("auto", () => {
await expect(handler(event(), context())).resolves.toBe("anthropic/claude-sonnet-5");
await expect(handler(event(), context())).resolves.toBe("anthropic/claude-sonnet-5");
- expect(doEvaluate).toHaveBeenCalledOnce();
+ expect(doDecide).toHaveBeenCalledOnce();
expect(runtime.logWarn).toHaveBeenCalledOnce();
- expect(runtime.logWarn).toHaveBeenCalledWith("model evaluation failed; using fallback", {
+ expect(runtime.logWarn).toHaveBeenCalledWith("model decision failed; using fallback", {
error: expect.objectContaining({
- message: expect.stringContaining("evaluation unavailable"),
+ message: expect.stringContaining("decision unavailable"),
}),
fallback: "anthropic/claude-sonnet-5",
turnId: "turn_1",
@@ -196,9 +196,9 @@ describe("auto", () => {
it("supports a provider model and reasoning as the fallback", async () => {
const fallback = anthropic("sonnet-5");
- const failed = new Experimental_EvaluationMockModelV4({
- doEvaluate: async () => {
- throw new Error("evaluation unavailable");
+ const failed = new Experimental_DecisionMockModelV4({
+ doDecide: async () => {
+ throw new Error("decision unavailable");
},
});
const handler = auto({
@@ -212,15 +212,15 @@ describe("auto", () => {
reasoning: "low",
});
expect(runtime.logWarn).toHaveBeenCalledWith(
- "model evaluation failed; using fallback",
+ "model decision failed; using fallback",
expect.objectContaining({ fallback: "anthropic.messages/sonnet-5" }),
);
});
it("propagates provider errors without a fallback and always propagates cancellation", async () => {
- const providerError = new Error("evaluation unavailable");
- const failed = new Experimental_EvaluationMockModelV4({
- doEvaluate: async () => {
+ const providerError = new Error("decision unavailable");
+ const failed = new Experimental_DecisionMockModelV4({
+ doDecide: async () => {
throw providerError;
},
});
@@ -229,8 +229,8 @@ describe("auto", () => {
runtime.state = new ContextContainer();
const controller = new AbortController();
- const pendingModel = new Experimental_EvaluationMockModelV4({
- doEvaluate: ({ abortSignal }) =>
+ const pendingModel = new Experimental_DecisionMockModelV4({
+ doDecide: ({ abortSignal }) =>
new Promise((_, reject) => {
abortSignal?.addEventListener("abort", () => reject(abortSignal.reason), { once: true });
}),
@@ -247,24 +247,24 @@ describe("auto", () => {
});
it("rejects invalid configurations and input", async () => {
- const evaluator = evaluationModel().model;
- expect(() => auto({ model: evaluator, options: {} })).toThrow("at least one option");
- expect(() => auto({ model: evaluator, options: { broken: "" } })).toThrow();
- expect(() => auto({ model: "", options })).toThrow("valid evaluation model");
- expect(() => auto({ model: evaluator, fallback: "", options })).toThrow("fallback model");
- expect(() => auto({ model: evaluator, fallback: {} as never, options })).toThrow(
+ const decider = decisionModel().model;
+ expect(() => auto({ model: decider, options: {} })).toThrow("at least one option");
+ expect(() => auto({ model: decider, options: { broken: "" } })).toThrow();
+ expect(() => auto({ model: "", options })).toThrow("valid decision model");
+ expect(() => auto({ model: decider, fallback: "", options })).toThrow("fallback model");
+ expect(() => auto({ model: decider, fallback: {} as never, options })).toThrow(
"fallback model",
);
expect(() =>
auto({
- model: evaluator,
+ model: decider,
options: {
broken: { model: "openai/large", description: "Difficult", reasoning: "maximum" },
},
} as never),
).toThrow();
- const handler = auto({ model: evaluator, options }).events["step.started"]!;
+ const handler = auto({ model: decider, options }).events["step.started"]!;
await expect(handler(event(), context(" "))).rejects.toThrow("requires user text");
});
});
diff --git a/packages/eve/src/models/auto.ts b/packages/eve/src/models/auto.ts
index d22f13714..3fc0aa33d 100644
--- a/packages/eve/src/models/auto.ts
+++ b/packages/eve/src/models/auto.ts
@@ -1,6 +1,6 @@
import { createHash } from "node:crypto";
-import { type Experimental_EvaluationModel as EvaluationModel } from "ai";
+import { type Experimental_DecisionModel as DecisionModel } from "ai";
import { loadContext } from "#context/container.js";
import { ContextKey } from "#context/key.js";
@@ -17,7 +17,7 @@ import type {
PublicAgentStaticModelDefinition,
} from "#shared/agent-definition.js";
-import { DEFAULT_EVALUATION_MODEL, evaluate } from "#ai/evaluate.js";
+import { DEFAULT_DECISION_MODEL, decide } from "#ai/decide.js";
type AutoModelSelection =
| PublicAgentStaticModelDefinition
@@ -37,9 +37,9 @@ type AutoOption =
interface AutoConfig<
T extends Readonly<Record<string, AutoOption>> = Readonly<Record<string, AutoOption>>,
> {
- /** Evaluation model instance or ID. Defaults to TypeSafe Jev through AI SDK model resolution. */
- readonly model?: EvaluationModel;
- /** Model selection to use when evaluation fails. */
+ /** Decision model instance or ID. Defaults to TypeSafe Jev through AI SDK model resolution. */
+ readonly model?: DecisionModel;
+ /** Model selection to use when decision fails. */
readonly fallback?: AutoModelSelection;
readonly options: T;
}
@@ -115,7 +115,7 @@ function turnId(event: unknown): string {
return event.data.turnId;
}
-function routingState(ctx: DynamicResolveContext): Parameters<typeof evaluate>[0]["state"] {
+function routingState(ctx: DynamicResolveContext): Parameters<typeof decide>[0]["state"] {
const messages: { role: string; text: string }[] = [];
let characters = 0;
@@ -146,7 +146,7 @@ function routingState(ctx: DynamicResolveContext): Parameters<typeof evaluate>[0
return { messages };
}
-/** Select a language model from the current prompt with an AI SDK evaluation model. */
+/** Select a language model from the current prompt with an AI SDK decision model. */
export function auto<const T extends Readonly<Record<string, AutoOption>>>(
config: AutoConfig<T>,
): DynamicSentinel<PublicAgentDynamicModelResult> {
@@ -168,11 +168,11 @@ export function auto<const T extends Readonly<Record<string, AutoOption>>>(
)
) {
throw new Error(
- "auto requires descriptions or { model, description, reasoning? } option entries and, when provided, a valid evaluation model and fallback model.",
+ "auto requires descriptions or { model, description, reasoning? } option entries and, when provided, a valid decision model and fallback model.",
);
}
- const evaluationModel = config.model ?? DEFAULT_EVALUATION_MODEL;
+ const decisionModel = config.model ?? DEFAULT_DECISION_MODEL;
const options = Object.entries(config.options).map(([key, option]) => ({
key,
model: typeof option === "string" ? key : option.model,
@@ -196,7 +196,7 @@ export function auto<const T extends Readonly<Record<string, AutoOption>>>(
const fingerprint = createHash("sha256")
.update(
JSON.stringify({
- evaluationModel: modelIdentity(evaluationModel),
+ decisionModel: modelIdentity(decisionModel),
fallback: config.fallback === undefined ? null : selectionIdentity(config.fallback),
options: options.map(({ key, model, description, reasoning }) => ({
key,
@@ -208,7 +208,7 @@ export function auto<const T extends Readonly<Record<string, AutoOption>>>(
)
.digest("hex");
const selection = new ContextKey<{ turnId: string; model: string }>(
- `eve.experimental.evaluate.model.${fingerprint}`,
+ `eve.experimental.decide.model.${fingerprint}`,
);
return defineDynamic({
@@ -220,11 +220,11 @@ export function auto<const T extends Readonly<Record<string, AutoOption>>>(
const previous = state.get(selection);
if (previous?.turnId === currentTurnId) return models.get(previous.model)!;
- const stateForEvaluation = routingState(ctx);
+ const stateForDecision = routingState(ctx);
try {
- const result = await evaluate({
- model: evaluationModel,
- state: stateForEvaluation,
+ const result = await decide({
+ model: decisionModel,
+ state: stateForDecision,
questions: {
route: {
type: "choice",
@@ -243,7 +243,7 @@ export function auto<const T extends Readonly<Record<string, AutoOption>>>(
} catch (error) {
ctx.abortSignal?.throwIfAborted();
if (config.fallback === undefined) throw error;
- log.warn("model evaluation failed; using fallback", {
+ log.warn("model decision failed; using fallback", {
error: formatError(error),
fallback: selectionLogIdentity(config.fallback),
turnId: currentTurnId,
diff --git a/packages/eve/src/public/ai/index.ts b/packages/eve/src/public/ai/index.ts
index a53ad5372..e9f0af6c6 100644
--- a/packages/eve/src/public/ai/index.ts
+++ b/packages/eve/src/public/ai/index.ts
@@ -1 +1 @@
-export { evaluate } from "#ai/evaluate.js";
+export { decide } from "#ai/decide.js";
diff --git a/packages/eve/src/public/definitions/connections/mcp.ts b/packages/eve/src/public/definitions/connections/mcp.ts
index b72424182..e9c714d10 100644
--- a/packages/eve/src/public/definitions/connections/mcp.ts
+++ b/packages/eve/src/public/definitions/connections/mcp.ts
@@ -72,7 +72,7 @@ export interface McpClientConnectionDefinition {
* - `never()`: allow all tool calls without approval
* - `once()`: require approval only the first time per session
* - `always()`: require approval for every tool call
- * - `auto()`: use an evaluation model to ask about dangerous or unclear effects
+ * - `auto()`: use a decision model to ask about dangerous or unclear effects
*
* When omitted, tool calls execute without approval, consistent
* with authored tools.
diff --git a/packages/eve/src/public/definitions/connections/openapi.ts b/packages/eve/src/public/definitions/connections/openapi.ts
index 8b3f4c6c8..ffa61fbb5 100644
--- a/packages/eve/src/public/definitions/connections/openapi.ts
+++ b/packages/eve/src/public/definitions/connections/openapi.ts
@@ -89,7 +89,7 @@ export interface OpenAPIConnectionDefinition {
* - `never()`: allow all tool calls without approval
* - `once()`: require approval only the first time per session
* - `always()`: require approval for every tool call
- * - `auto()`: use an evaluation model to ask about dangerous or unclear effects
+ * - `auto()`: use a decision model to ask about dangerous or unclear effects
*/
approval?: Approval;
/**
diff --git a/packages/eve/src/public/models/openai/chatgpt/model.test.ts b/packages/eve/src/public/models/openai/chatgpt/model.test.ts
index 9f4b225e0..e6487b594 100644
--- a/packages/eve/src/public/models/openai/chatgpt/model.test.ts
+++ b/packages/eve/src/public/models/openai/chatgpt/model.test.ts
@@ -69,7 +69,7 @@ describe("Codex model", () => {
{ role: "user", content: [{ type: "input_text", text: "hello" }] },
{
role: "assistant",
- content: [{ type: "output_text", text: "previous answer" }],
+ content: "previous answer",
phase: "final_answer",
},
]);
diff --git a/packages/eve/src/tools/approval/policies.test.ts b/packages/eve/src/tools/approval/policies.test.ts
index b3c05bccf..6098dfe4e 100644
--- a/packages/eve/src/tools/approval/policies.test.ts
+++ b/packages/eve/src/tools/approval/policies.test.ts
@@ -1,7 +1,7 @@
import { beforeEach, describe, expect, it, vi } from "vitest";
-const evaluate = vi.hoisted(() => vi.fn());
-vi.mock("#ai/evaluate.js", () => ({ evaluate }));
+const decide = vi.hoisted(() => vi.fn());
+vi.mock("#ai/decide.js", () => ({ decide }));
import type { ApprovalContext } from "#approval/definition.js";
import { always, auto, never, once } from "#tools/approval/policies.js";
@@ -25,7 +25,7 @@ function approvalContext(overrides: Partial<ApprovalContext> = {}): ApprovalCont
}
describe("dynamic tool approval helpers", () => {
- beforeEach(() => evaluate.mockReset());
+ beforeEach(() => decide.mockReset());
it.each([
["always", always(), "user-approval"],
@@ -42,11 +42,11 @@ describe("dynamic tool approval helpers", () => {
["clear", "approved"],
["caution", "user-approval"],
] as const)("maps the automatic review's %s decision to %s", async (choice, expected) => {
- evaluate.mockResolvedValueOnce({ answers: { permission: { choice } } });
+ decide.mockResolvedValueOnce({ answers: { permission: { choice } } });
const context = approvalContext();
await expect(auto()(context)).resolves.toBe(expected);
- expect(evaluate).toHaveBeenCalledWith(
+ expect(decide).toHaveBeenCalledWith(
expect.objectContaining({
abortSignal: expect.any(AbortSignal),
maxRetries: 0,
@@ -57,11 +57,11 @@ describe("dynamic tool approval helpers", () => {
});
it("allows classifier instructions and outcome descriptions to be overridden", async () => {
- evaluate.mockResolvedValueOnce({ answers: { permission: { choice: "clear" } } });
+ decide.mockResolvedValueOnce({ answers: { permission: { choice: "clear" } } });
const context = approvalContext();
await auto({
- model: "custom-evaluator",
+ model: "custom-decider",
instructions: "Classify this action for my application.",
criteria: {
clear: "The action is within policy.",
@@ -69,9 +69,9 @@ describe("dynamic tool approval helpers", () => {
},
})(context);
- expect(evaluate).toHaveBeenCalledWith(
+ expect(decide).toHaveBeenCalledWith(
expect.objectContaining({
- model: "custom-evaluator",
+ model: "custom-decider",
questions: {
permission: {
type: "choice",
@@ -86,40 +86,40 @@ describe("dynamic tool approval helpers", () => {
);
});
- it("accepts another evaluation model", async () => {
- evaluate.mockResolvedValueOnce({ answers: { permission: { choice: "clear" } } });
+ it("accepts another decision model", async () => {
+ decide.mockResolvedValueOnce({ answers: { permission: { choice: "clear" } } });
const model = { modelId: "custom", provider: "test", specificationVersion: "v4" } as never;
const context = approvalContext();
await expect(auto({ model })(context)).resolves.toBe("approved");
- expect(evaluate).toHaveBeenCalledWith(expect.objectContaining({ model }));
+ expect(decide).toHaveBeenCalledWith(expect.objectContaining({ model }));
});
it.each([new Date(), new Map([["kind", "delete"]]), new Set(["delete"]), Number.NaN])(
- "requires approval without evaluating lossy input %#",
+ "requires approval without deciding lossy input %#",
async (toolInput) => {
await expect(auto()(approvalContext({ toolInput: { value: toolInput } }))).resolves.toBe(
"user-approval",
);
- expect(evaluate).not.toHaveBeenCalled();
+ expect(decide).not.toHaveBeenCalled();
},
);
- it("requires approval without evaluating oversized input", async () => {
+ it("requires approval without deciding oversized input", async () => {
const context = approvalContext({ toolInput: { value: "x".repeat(64 * 1024) } });
await expect(auto()(context)).resolves.toBe("user-approval");
- expect(evaluate).not.toHaveBeenCalled();
+ expect(decide).not.toHaveBeenCalled();
});
it("fails closed when the automatic review is unavailable", async () => {
- evaluate.mockRejectedValueOnce(new Error("unavailable"));
+ decide.mockRejectedValueOnce(new Error("unavailable"));
await expect(auto()(approvalContext())).resolves.toBe("user-approval");
});
it("propagates cancellation instead of requesting approval", async () => {
const reason = new Error("cancelled");
- evaluate.mockRejectedValueOnce(reason);
+ decide.mockRejectedValueOnce(reason);
const context = approvalContext({ abortSignal: AbortSignal.abort(reason) });
await expect(auto()(context)).rejects.toBe(reason);
diff --git a/packages/eve/src/tools/approval/policies.ts b/packages/eve/src/tools/approval/policies.ts
index 888f61591..6a0ac0edc 100644
--- a/packages/eve/src/tools/approval/policies.ts
+++ b/packages/eve/src/tools/approval/policies.ts
@@ -1,7 +1,7 @@
-import type { Experimental_EvaluationModel as EvaluationModel } from "ai";
+import type { Experimental_DecisionModel as DecisionModel } from "ai";
import type { ApprovalContext, ApprovalPolicy } from "#approval/definition.js";
-import { evaluate } from "#ai/evaluate.js";
+import { decide } from "#ai/decide.js";
import { parseJsonValue, type JsonObject } from "#shared/json.js";
import { stampDurableDynamicCallback } from "#tools/durable-callbacks.js";
@@ -12,8 +12,8 @@ const DEFAULT_CLEAR_DESCRIPTION = "The action is routine and low impact.";
const DEFAULT_CAUTION_DESCRIPTION = "The action is dangerous or its important effects are unclear.";
export interface AutoApprovalOptions {
- /** Evaluation model instance or ID. Defaults to TypeSafe Jev. */
- readonly model?: EvaluationModel;
+ /** Decision model instance or ID. Defaults to TypeSafe Jev. */
+ readonly model?: DecisionModel;
/** Instructions used to classify the exact tool call. */
readonly instructions?: string;
/** Descriptions of the fixed approval outcomes. */
@@ -24,7 +24,7 @@ export interface AutoApprovalOptions {
}
async function autoApproval(
- evaluationModel: EvaluationModel,
+ decisionModel: DecisionModel,
options: AutoApprovalOptions,
context: ApprovalContext,
): Promise<"approved" | "user-approval"> {
@@ -35,8 +35,8 @@ async function autoApproval(
return "user-approval";
}
const abortSignal = AbortSignal.any([context.abortSignal, AbortSignal.timeout(30_000)]);
- const result = await evaluate({
- model: evaluationModel,
+ const result = await decide({
+ model: decisionModel,
state: {
action: {
input,
@@ -79,16 +79,16 @@ function onceApproval(
}
/**
- * Returns an `approval` callback that asks an AI SDK evaluation model whether
+ * Returns an `approval` callback that asks an AI SDK decision model whether
* the exact tool call presents concrete security danger. Defaults to TypeSafe Jev.
- * Unavailable or invalid evaluations require user approval.
+ * Unavailable or invalid decisions require user approval.
*/
export function auto<TInput = unknown>(options: AutoApprovalOptions = {}): ApprovalPolicy<TInput> {
- const evaluationModel = options.model ?? "typesafe-ai/jev";
+ const decisionModel = options.model ?? "typesafe-ai/jev";
const callback = (_closure: JsonObject, context: ApprovalContext) =>
- autoApproval(evaluationModel, options, context);
+ autoApproval(decisionModel, options, context);
return stampDurableDynamicCallback(
- (context) => autoApproval(evaluationModel, options, context as ApprovalContext),
+ (context) => autoApproval(decisionModel, options, context as ApprovalContext),
{ callback, closure: {} },
) as ApprovalPolicy<TInput>;
}
diff --git a/packages/eve/src/tools/provided/agent-router.test.ts b/packages/eve/src/tools/provided/agent-router.test.ts
index 87bb8f9d2..2ab8b393a 100644
--- a/packages/eve/src/tools/provided/agent-router.test.ts
+++ b/packages/eve/src/tools/provided/agent-router.test.ts
@@ -2,10 +2,10 @@ import { beforeEach, describe, expect, it, vi } from "vitest";
import { agentRouter } from "#tools/provided/agent-router.js";
import { runAgentRouterTask } from "#execution/tools/agent-router.js";
-import { evaluate } from "#ai/evaluate.js";
+import { decide } from "#ai/decide.js";
import type { WorkflowToolContext } from "#tools/workflow-definition.js";
-vi.mock("#ai/evaluate.js", () => ({ evaluate: vi.fn() }));
+vi.mock("#ai/decide.js", () => ({ decide: vi.fn() }));
describe("agentRouter", () => {
beforeEach(() => vi.clearAllMocks());
@@ -19,7 +19,7 @@ describe("agentRouter", () => {
});
it("routes through all workflow agent descriptions", async () => {
- vi.mocked(evaluate).mockResolvedValue({
+ vi.mocked(decide).mockResolvedValue({
answers: { route: { choice: "operator", type: "choice" } },
} as never);
const { agent, send } = replyingAgent("operated");
@@ -36,7 +36,7 @@ describe("agentRouter", () => {
await expect(runAgentRouterTask({ message: "Deploy the service" }, ctx)).resolves.toBe(
"operated",
);
- expect(evaluate).toHaveBeenCalledWith({
+ expect(decide).toHaveBeenCalledWith({
abortSignal,
state: { message: "Deploy the service" },
questions: {
@@ -65,12 +65,12 @@ describe("agentRouter", () => {
});
await expect(runAgentRouterTask({ message: "Investigate" }, ctx)).resolves.toBe("researched");
- expect(evaluate).not.toHaveBeenCalled();
+ expect(decide).not.toHaveBeenCalled();
expect(agent).toHaveBeenCalledWith("researcher");
expect(send).toHaveBeenCalledWith("Investigate", { signal: expect.any(AbortSignal) });
});
- it("invokes the only available described agent without evaluation", async () => {
+ it("invokes the only available described agent without decision", async () => {
const { agent, send } = replyingAgent("researched");
const ctx = workflowContext({
agent,
@@ -78,12 +78,12 @@ describe("agentRouter", () => {
});
await expect(runAgentRouterTask({ message: "Investigate" }, ctx)).resolves.toBe("researched");
- expect(evaluate).not.toHaveBeenCalled();
+ expect(decide).not.toHaveBeenCalled();
expect(agent).toHaveBeenCalledWith("researcher");
expect(send).toHaveBeenCalledWith("Investigate", { signal: expect.any(AbortSignal) });
});
- it("rejects an agent map without descriptions before evaluation", async () => {
+ it("rejects an agent map without descriptions before decision", async () => {
const ctx = workflowContext({
agent: vi.fn(),
agents: { agent: { description: " " } },
@@ -92,7 +92,7 @@ describe("agentRouter", () => {
await expect(runAgentRouterTask({ message: "Route me" }, ctx)).rejects.toThrow(
"agentRouter requires at least one available agent with a description.",
);
- expect(evaluate).not.toHaveBeenCalled();
+ expect(decide).not.toHaveBeenCalled();
});
});
diff --git a/pnpm-lock.yaml b/pnpm-lock.yaml
index 64435d8bb..86d3a485f 100644
--- a/pnpm-lock.yaml
+++ b/pnpm-lock.yaml
@@ -165,29 +165,29 @@ settings:
catalogs:
default:
'@ai-sdk/anthropic':
- specifier: ^4.0.49
- version: 4.0.56
+ specifier: ^4.0.72
+ version: 4.0.72
'@ai-sdk/code-mode':
- specifier: ^1.0.62
- version: 1.0.62
+ specifier: ^1.0.85
+ version: 1.0.85
'@ai-sdk/google':
- specifier: ^4.0.64
- version: 4.0.74
+ specifier: ^4.0.88
+ version: 4.0.88
'@ai-sdk/mcp':
- specifier: ^2.0.45
- version: 2.0.52
+ specifier: ^2.0.67
+ version: 2.0.67
'@ai-sdk/openai':
- specifier: ^4.0.60
- version: 4.0.69
+ specifier: ^4.0.84
+ version: 4.0.84
'@ai-sdk/otel':
- specifier: ^1.0.93
- version: 1.0.105
+ specifier: ^1.0.128
+ version: 1.0.128
'@ai-sdk/provider':
- specifier: ^4.0.10
- version: 4.0.21
+ specifier: ^4.0.22
+ version: 4.0.22
'@ai-sdk/provider-utils':
- specifier: ^5.0.36
- version: 5.0.53
+ specifier: ^5.0.54
+ version: 5.0.54
'@opentelemetry/context-async-hooks':
specifier: 2.6.1
version: 2.6.1
@@ -219,8 +219,8 @@ catalogs:
specifier: 5.0.1
version: 5.0.1
ai:
- specifier: ^7.0.116
- version: 7.0.127
+ specifier: ^7.0.128
+ version: 7.0.128
marked:
specifier: 17.0.6
version: 17.0.6
@@ -319,13 +319,13 @@ importers:
version: 1.1.1([email protected])
'@vercel/analytics':
specifier: 2.0.1
- version: 2.0.1(e9f01ee4848ee8142aacdb51dfc5a6a4)
'@vercel/geistdocs':
specifier: 2.9.0
version: 2.9.0(79560e942e5066c3fa0e250d382bc1cd)
'@vercel/speed-insights':
specifier: 2.0.0
- version: 2.0.0(e9f01ee4848ee8142aacdb51dfc5a6a4)
'@vgpu/core':
specifier: 0.0.7
version: 0.0.7
@@ -392,10 +392,10 @@ importers:
version: 0.33.2(@vercel/[email protected])(eve@packages+eve)
'@agentphone/chat-sdk-adapter':
specifier: 0.1.0
'@beeper/chat-adapter-matrix':
specifier: 0.2.0
'@blitzreels/eve':
specifier: ^0.2.1
version: 0.2.1(eve@packages+eve)
@@ -404,55 +404,55 @@ importers:
version: 0.1.0(@cfworker/[email protected])(eve@packages+eve)([email protected])([email protected])
'@chat-adapter/gchat':
specifier: 4.41.0
'@chat-adapter/gmail':
specifier: 4.41.0
'@chat-adapter/messenger':
specifier: 4.41.0
'@chat-adapter/state-memory':
specifier: 4.41.0
'@chat-adapter/state-redis':
specifier: 4.41.0
'@chat-adapter/whatsapp':
specifier: 4.41.0
'@chat-adapter/x':
specifier: 4.41.0
'@getdial/chat-sdk-adapter':
specifier: 0.2.0
'@github-tools/eve-extension':
specifier: ^0.1.0
'@jetty/eve':
specifier: ^0.3.0
version: 0.3.1(eve@packages+eve)
'@kapso/chat-adapter':
specifier: 0.1.1
'@kybernesis/arcana':
specifier: 0.4.1
version: 0.4.1(eve@packages+eve)
'@larksuite/vercel-chat-adapter':
specifier: 0.1.2
'@liveblocks/chat-sdk-adapter':
specifier: 3.23.0
'@mdx-js/mdx':
specifier: 3.1.1
version: 3.1.1([email protected])
'@novu/chat-sdk-adapter':
specifier: 0.1.0
'@onkernel/eve-extension':
specifier: ^0.1.3
'@opentelemetry/api':
specifier: 1.9.1
version: 1.9.1
@@ -467,7 +467,7 @@ importers:
'@resend/chat-sdk-adapter':
specifier: 0.2.2
'@stripe/link-integrations-eve':
specifier: ^0.2.0
version: 0.2.0(eve@packages+eve)
@@ -497,7 +497,7 @@ importers:
version: 0.184.1
'@upstash/agentkit-eve':
specifier: 0.9.0
- version: 0.9.0(@upstash/[email protected])([email protected]([email protected]))(eve@packages+eve)
+ version: 0.9.0(@upstash/[email protected])([email protected]([email protected]))(eve@packages+eve)
'@upstash/redis':
specifier: ^1.38.4
version: 1.38.4
@@ -506,10 +506,10 @@ importers:
version: 0.2.1(eve@packages+eve)
'@veltdev/chat-sdk-adapter':
specifier: 0.2.4
'@vercel/connect':
specifier: 2.2.0
'@vercel/otel':
specifier: 'catalog:'
version: 2.1.3(@opentelemetry/[email protected])(@opentelemetry/[email protected])(@opentelemetry/[email protected](@opentelemetry/[email protected])([email protected]))(@opentelemetry/[email protected](@opentelemetry/[email protected]))(@opentelemetry/[email protected](@opentelemetry/[email protected]))(@opentelemetry/[email protected](@opentelemetry/[email protected]))(@opentelemetry/[email protected](@opentelemetry/[email protected]))
@@ -521,22 +521,22 @@ importers:
version: 0.1.72
'@zernio/chat-sdk-adapter':
specifier: 0.5.0
braintrust:
specifier: 3.29.0
chat:
specifier: 4.41.0
chat-adapter-sendblue:
specifier: 0.2.0
eve:
specifier: workspace:*
version: link:../../packages/eve
eve-channel-blooio:
specifier: 0.2.0
- version: 0.2.0([email protected]([email protected]))(eve@packages+eve)
+ version: 0.2.0([email protected]([email protected]))(eve@packages+eve)
mdast-util-mdx-jsx:
specifier: 3.2.0
version: 3.2.0([email protected])
@@ -612,10 +612,10 @@ importers:
version: 0.41.2
'@vercel/connect':
specifier: 2.2.0
ai:
specifier: 'catalog:'
- version: 7.0.127([email protected])
+ version: 7.0.128([email protected])
eve:
specifier: workspace:*
version: link:../../../packages/eve
@@ -664,7 +664,7 @@ importers:
dependencies:
ai:
specifier: 'catalog:'
- version: 7.0.127([email protected])
+ version: 7.0.128([email protected])
eve:
specifier: workspace:*
version: link:../../../packages/eve
@@ -949,7 +949,7 @@ importers:
specifier: 'catalog:'
version: 7.0.2
- e2e/fixtures/agent-dynamic-model-session:
+ e2e/fixtures/agent-decide:
dependencies:
'@eve-e2e/config':
specifier: workspace:*
@@ -957,12 +957,12 @@ importers:
'@workflow/world-postgres':
specifier: 'catalog:'
+ ai:
+ specifier: 'catalog:'
+ version: 7.0.128([email protected])
eve:
specifier: workspace:*
version: link:../../../packages/eve
- zod:
- specifier: 'catalog:'
- version: 4.5.4
devDependencies:
'@types/node':
specifier: 'catalog:'
@@ -971,7 +971,7 @@ importers:
specifier: 'catalog:'
version: 7.0.2
- e2e/fixtures/agent-evaluate:
+ e2e/fixtures/agent-dynamic-model-session:
dependencies:
'@eve-e2e/config':
specifier: workspace:*
@@ -979,12 +979,12 @@ importers:
'@workflow/world-postgres':
specifier: 'catalog:'
- ai:
- specifier: 'catalog:'
- version: 7.0.127([email protected])
eve:
specifier: workspace:*
version: link:../../../packages/eve
+ zod:
+ specifier: 'catalog:'
+ version: 4.5.4
devDependencies:
'@types/node':
specifier: 'catalog:'
@@ -1192,7 +1192,7 @@ importers:
version: link:../e2e-config
'@vercel/connect':
specifier: 2.2.0
'@workflow/world-postgres':
specifier: 'catalog:'
@@ -1503,7 +1503,7 @@ importers:
dependencies:
'@ai-sdk/openai':
specifier: 'catalog:'
- version: 4.0.69([email protected])
+ version: 4.0.84([email protected])
'@opentelemetry/sdk-trace-base':
specifier: 'catalog:'
version: 2.6.1(@opentelemetry/[email protected])
@@ -1639,40 +1639,40 @@ importers:
version: 1.3.0([email protected])
'@ai-sdk/anthropic':
specifier: 'catalog:'
- version: 4.0.56([email protected])
+ version: 4.0.72([email protected])
'@ai-sdk/code-mode':
specifier: 'catalog:'
'@ai-sdk/google':
specifier: 'catalog:'
- version: 4.0.74([email protected])
+ version: 4.0.88([email protected])
'@ai-sdk/mcp':
specifier: 'catalog:'
- version: 2.0.52([email protected])
+ version: 2.0.67([email protected])
'@ai-sdk/openai':
specifier: 'catalog:'
- version: 4.0.69([email protected])
+ version: 4.0.84([email protected])
'@ai-sdk/otel':
specifier: 'catalog:'
- version: 1.0.105([email protected])
+ version: 1.0.128([email protected])
'@ai-sdk/provider':
specifier: 'catalog:'
- version: 4.0.21
+ version: 4.0.22
'@ai-sdk/provider-utils':
specifier: 'catalog:'
- version: 5.0.53([email protected])
+ version: 5.0.54([email protected])
'@cfworker/json-schema':
specifier: 4.1.1
version: 4.1.1
'@chat-adapter/slack':
specifier: 4.41.0
'@chat-adapter/state-memory':
specifier: 4.41.0
'@chat-adapter/twilio':
specifier: 4.41.0
'@clack/core':
specifier: 1.3.1
version: 1.3.1
@@ -1681,7 +1681,7 @@ importers:
version: link:../eve-catalog
'@linqapp/chat-sdk-adapter':
specifier: ^0.5.1
'@modelcontextprotocol/server':
specifier: 2.0.0
version: 2.0.0
@@ -1705,7 +1705,7 @@ importers:
version: 2.6.1(@opentelemetry/[email protected])
'@photon-ai/chat-adapter-imessage':
specifier: 3.2.0
'@slack/web-api':
specifier: 7.19.0
@@ -1729,7 +1729,7 @@ importers:
version: 2.8.0
'@vercel/connect':
specifier: 'catalog:'
'@vercel/detect-agent':
specifier: 1.2.3
version: 1.2.3
@@ -1765,10 +1765,10 @@ importers:
version: 5.0.1(@aws-sdk/[email protected])(@opentelemetry/[email protected])([email protected])
ai:
specifier: 'catalog:'
- version: 7.0.127([email protected])
+ version: 7.0.128([email protected])
chat:
specifier: 4.41.0
chokidar:
specifier: 5.0.0
version: 5.0.0
@@ -1874,7 +1874,7 @@ importers:
dependencies:
'@vercel/connect':
specifier: 'catalog:'
eve:
specifier: workspace:*
version: link:../eve
@@ -1966,8 +1966,8 @@ packages:
peerDependencies:
zod: ^3.25.76 || ^4.1.8
- '@ai-sdk/[email protected]':
- resolution: {integrity: sha512-rpz9LepVvjCs0YXWN/I5DLNAK/SbaA8g2sMTS1iyWQyH7Vsjqeonc3A7WbZiM6bfhNtrGO6irhvkDjL/xWpE1w==}
+ '@ai-sdk/[email protected]':
+ resolution: {integrity: sha512-dN5ShH8f6MWlSBSx4BdL+/cfPvByFYMo2Hr9UB6XeHY7/9tbcYU/ZukptBcyWJzqMsQOl3gSfg99iWdyNuH3kg==}
engines: {node: '>=22'}
peerDependencies:
zod: ^3.25.76 || ^4.1.8
@@ -1984,11 +1984,11 @@ packages:
peerDependencies:
zod: ^3.25.76 || ^4.1.8
- '@ai-sdk/[email protected]':
- resolution: {integrity: sha512-Pk6lZIJgq5n6O8hrPFm0XLFPGvA8f1Yew9d7+GSXgn1ajlXy/Uqn/8j3o5JuelstFCS2OJmgG4vtiZrsBTKE1Q==}
+ '@ai-sdk/[email protected]':
+ resolution: {integrity: sha512-GBX2uumgOLsA4OeDYSoiN3ET4R0Y5KvzlIEUr0ampZRLqpsPyPhFv+YjVJSIcyw9i7+t2Vlieg9AmjuYfuVV4Q==}
engines: {node: '>=22.13.0'}
peerDependencies:
- ai: 7.0.105
+ ai: 7.0.128
'@ai-sdk/[email protected]':
resolution: {integrity: sha512-sF+t6MnK1PTNN1oqmJpnWCm/4reVOpKryR1EcIl17O5Bu+AFstD/DpFZueRtzFk1dEZ0+uPX6gZXndjrNPAxLQ==}
@@ -1996,8 +1996,8 @@ packages:
peerDependencies:
zod: ^3.25.76 || ^4.1.8
- '@ai-sdk/[email protected]':
- resolution: {integrity: sha512-USFJ7MIRxR6+PrXT06Q7V+cJf/rJIOmH/KdY84OWCb6+9bhg4xg1pUNxKD28N4F8+VFnX42wfWtqbM3+ZhK+yQ==}
+ '@ai-sdk/[email protected]':
+ resolution: {integrity: sha512-fPkB0RKy2ZD9yxC/3XeZR/9d7J/qijsix8reBPgEtiUCnukr0XQ/3TaH+UGa0U9CJyhrJENquRGyi83f+FGwiA==}
engines: {node: '>=18'}
peerDependencies:
zod: ^3.25.76 || ^4.1.8
@@ -2008,14 +2008,14 @@ packages:
peerDependencies:
zod: ^3.25.76 || ^4.1.8
- '@ai-sdk/[email protected]':
- resolution: {integrity: sha512-nnGTUHzPRQ14jhv9CL4OQrzcRBIFTBMIPKGIi9Z9u/4GtwSRxa/nxUX6np2Iwp5phOzO0a0njj/j9Jolf5AEaA==}
- engines: {node: '>=22'}
+ '@ai-sdk/[email protected]':
+ resolution: {integrity: sha512-XUC2HGvDLhpfHKT7yXirZELHWIHzPJLtE3CE3FO6fjcQHcKR5nrsHxOp00THk1SivDo3IZB1yS6Un4+kAQ8eJg==}
+ engines: {node: '>=18'}
peerDependencies:
zod: ^3.25.76 || ^4.1.8
- '@ai-sdk/[email protected]':
- resolution: {integrity: sha512-c8ztST/CslupqO3p7izsAuG3TlKg9mhE6mGJ42iUADsSMJIV5IOKd9AsLG33QCAY7SWs6NDO2giM/rq7bUTNlQ==}
+ '@ai-sdk/[email protected]':
+ resolution: {integrity: sha512-4lTyt8a4BMfVz4QGo+VO5odakmYMX5kq+xiRgdQGGY7o/jP2WSasp6LvIvYvt/lHzz1MfE+xKTOJhz7mMwjynA==}
engines: {node: '>=22'}
peerDependencies:
zod: ^3.25.76 || ^4.1.8
@@ -2032,8 +2032,8 @@ packages:
peerDependencies:
zod: ^3.25.76 || ^4.1.8
- '@ai-sdk/[email protected]':
- resolution: {integrity: sha512-MoWPhC+kwqiAsFF21iPDCxZPXW6lQYY1ImjohQ0q7VHgY4e8I/mfgKeeessYFI/mh0e7vSxG0/P767PSsTdngg==}
+ '@ai-sdk/[email protected]':
+ resolution: {integrity: sha512-5wfpAcnXrR2o1kUZURXhl6ErOGDmdayCi+GsfsB3vJHeITpm1t9aVqrw9/wUhpUmZTvvC5/+vDRtdo5QU9XYJA==}
engines: {node: '>=22'}
peerDependencies:
zod: ^3.25.76 || ^4.1.8
@@ -2044,8 +2044,8 @@ packages:
peerDependencies:
zod: ^3.25.76 || ^4.1.8
- '@ai-sdk/[email protected]':
- resolution: {integrity: sha512-01JlPlhpORspKA2xAh11qQ6KNieodbBEIJqMTzjmlV3mx6Y3JhBR4Uk8a7SsAv5ijphAUNZNYh50F2EhEEJ7fA==}
+ '@ai-sdk/[email protected]':
+ resolution: {integrity: sha512-cMEif3Lbvh+FozPpBM5UPy6QQhH8v8L/u1kcR2UXp+86EfIDXIaIeOaH+wYCk2CfloFccUn1K8LJuPXqZMZsfw==}
engines: {node: '>=22'}
peerDependencies:
zod: ^3.25.76 || ^4.1.8
@@ -2068,14 +2068,14 @@ packages:
peerDependencies:
zod: ^3.25.76 || ^4.1.8
- '@ai-sdk/[email protected]':
- resolution: {integrity: sha512-HAKua/QVNjwsBpJE3SUGLEPCSRThOLiiOy92gsUOkX4ZnEJHubkR6REHvi4RGGIn97lYBma+mq3BjBqgyNuhwQ==}
+ '@ai-sdk/[email protected]':
+ resolution: {integrity: sha512-RsHz+jqstQ+wnMex1BLzIlwseDRzJAruT7ZW1X1azCEI9cVkSQVn17JS6qVntdi+DKkAMPg4rP1APixDEZFKiw==}
engines: {node: '>=22'}
peerDependencies:
zod: ^3.25.76 || ^4.1.8
- '@ai-sdk/[email protected]':
- resolution: {integrity: sha512-n73OvB0nDgPXx9+3JwFO+/+WlBOQTIJBKD+vYTsmPoacyxjjQgaA2kqG4YUmdTkg+c/aXoeo07LLTayBLpz3KA==}
+ '@ai-sdk/[email protected]':
+ resolution: {integrity: sha512-WighejgfttBLk6VoI4Te2hOJmssvOGMCpKyVPrg4GVa08PYEuLZDBetDC+1OXt8mvodVYPm6CNEZ5FA+fJN7bA==}
engines: {node: '>=22'}
'@ai-sdk/[email protected]':
@@ -2096,20 +2096,26 @@ packages:
peerDependencies:
zod: ^3.25.76 || ^4.1.8
+ '@ai-sdk/[email protected]':
+ resolution: {integrity: sha512-xpURVGQud+AEOZ/uRZTBU0eVy2+dt/y/dUEp9ewjW3eTM6yu0vThzTvduhOSabpn6AExewg8WJW04P4D6USVXw==}
+ engines: {node: '>=18'}
+ peerDependencies:
+ zod: ^3.25.76 || ^4.1.8
+
'@ai-sdk/[email protected]':
resolution: {integrity: sha512-ukLTs9x1Xm6lxSIwbJIxYQxbZHmVeIczNntARZrBcOa6pjdBhqeZnATNnJMHa7M9dZ8Ji5NJbpKpvz7o5XyBtg==}
engines: {node: '>=18.17'}
peerDependencies:
zod: ^3.25.76 || ^4.1.8
- '@ai-sdk/[email protected]':
- resolution: {integrity: sha512-gw/bcNseOGSs59TMtV4H1KwqXXe24NHgx+uBYr98pa4Fg6Uvp8hPPxjuckVnfFFyIxHCDew8so0ujpvcSuyMZA==}
- engines: {node: '>=22'}
+ '@ai-sdk/[email protected]':
+ resolution: {integrity: sha512-89a7sPvZqXsP3PigcAQLWntu/duIVEcN/RyChAB+j6p3C7NwW767sVttvL8r0+BetTg63uAXQoYHCZZueQKBSw==}
+ engines: {node: '>=18.17'}
peerDependencies:
zod: ^3.25.76 || ^4.1.8
- '@ai-sdk/[email protected]':
- resolution: {integrity: sha512-VVe6UDd0y0/B4TfauuKKzcppqvhTmSluwwUVT4WG2/8h7gFbyFm7PM4UGM+ZXe3/RI6bv+Vl1cw/7aiBoQqdxg==}
+ '@ai-sdk/[email protected]':
+ resolution: {integrity: sha512-amqDxFnw9+dpVrDACRDGmExUUNOH9C1D2bhg/DqWJnamEOieDTXBBXLHujSby/0MqHdAFksnpE8Pzfqx6ABMBw==}
engines: {node: '>=22'}
peerDependencies:
zod: ^3.25.76 || ^4.1.8
@@ -2122,16 +2128,20 @@ packages:
resolution: {integrity: sha512-B0M50w0W43jTzZGyZrNwwaK1M79aKPw6PD/dQhkBR7vkR68yCINWBo2CbBLJrbrqYc46TaiSfsTiT+S9EmVl0w==}
engines: {node: '>=18'}
+ '@ai-sdk/[email protected]':
+ resolution: {integrity: sha512-cmG40u10QFEUrYXWYom6Lo6DLR+pND3wWABTTT9Nf7zg9F7BD46KNj+STekndp9Kg5ZAIPA0WxmTI0IwRvKvvQ==}
+ engines: {node: '>=18'}
+
'@ai-sdk/[email protected]':
resolution: {integrity: sha512-9Av6kg0t/IN/dcYAAmEJ4B9OPhcEJqwSR+GfHEA8olRkindXpUHadc3p3cyvgFUQFTVm1G5thtsmgZ9yVb2w3A==}
engines: {node: '>=18'}
- '@ai-sdk/[email protected]':
- resolution: {integrity: sha512-VYMBxIQdcHqbIf1j+YZlI9Ati6LZ4wJe0GGd4z4a5H/KxTggjeOiyaVYTnfF7LHZK5jMQ+rofmzz4QPqf++NUw==}
- engines: {node: '>=22'}
+ '@ai-sdk/[email protected]':
+ resolution: {integrity: sha512-IpefF5ssZVOZCD4Wu7zxt6PwZoGT2HLVjxpJ10Bjx6EAmRfLYJPjyrfxb0RYDlwvXnDb9nkQ7Do/oJVDuRFzew==}
+ engines: {node: '>=18'}
- '@ai-sdk/[email protected]':
- resolution: {integrity: sha512-UpbC9C1oht8dhfKPbVXSLRZS3DI8uk8n5v2uMjBPNIYGsC2kL045ywq7D8Hh9KgnH9rP5W/GxQdYtWScRouqlA==}
+ '@ai-sdk/[email protected]':
+ resolution: {integrity: sha512-1Jtmn36VNMOBo9CMUQrPqsR744qtqRloEiXpWCYS8ffPlhS2CrSNmoho5lPjp3R7EnkzBsGI6p51TA7fQJQZdA==}
engines: {node: '>=22'}
'@ai-sdk/[email protected]':
@@ -7876,8 +7886,8 @@ packages:
resolution: {integrity: sha512-kja8j7PjmncONqaTsB8fQ+wE2mSU2DJ9D4XKoJ5PFWIdRMa6SLSN1ff4mOr4jCbfRSsxR4keIiySJU0N9T5hIQ==}
engines: {node: '>= 8.0.0'}
- resolution: {integrity: sha512-Er9lVNxOmW3dwWg+s6/OZXyNZr36hA+SXbr/5p+OJtVVaNOkXrPEeX0rYWVpyMLsteefXrTcwFYQ0NvRjI3nlA==}
+ resolution: {integrity: sha512-MsutXPJ3G/mljr9pTb5xh3Ren8ZjRCm7EgqZoEHKfN7mOfzFOvac00hBnx5u/wDZg29P9mVoNN9Ro5C6DR4hKA==}
engines: {node: '>=18'}
peerDependencies:
zod: ^3.25.76 || ^4.1.8
@@ -7888,14 +7898,14 @@ packages:
peerDependencies:
zod: ^3.25.76 || ^4.1.8
- resolution: {integrity: sha512-gkV3W+tTtWwQ41KNV/KjV8wabBrk+bnbnZ3f0AWp/T9LLyljose8uhFViKleVy7tIdEiR2PiS4BpbNdcExf7rQ==}
- engines: {node: '>=22'}
+ resolution: {integrity: sha512-pb7vIIBXKw4VGpbLmXvD+u5qbUe+2Mm7hEEkHs6Z2XBiifxAUEB38gl5JbaRi3BWsFqN13GP25/NHhf/DvPjgg==}
+ engines: {node: '>=18'}
peerDependencies:
zod: ^3.25.76 || ^4.1.8
- resolution: {integrity: sha512-JNsNPk4ZvRGEJuqdVo2lYFkm7uAuf3p/RwKFWn1Wh/+IK3U1RyFWw2GXoFnhzqNfxEuj7XJB4WpIUhl1sHToWg==}
+ resolution: {integrity: sha512-WMKN83DodYtH52ETPHq0jrSwj3wU0aHkhLHXMrwMRabC/aMpIqeWsXJgjkUwrOqkaKHWONdupk/2ttPRvAyFcg==}
engines: {node: '>=22'}
peerDependencies:
zod: ^3.25.76 || ^4.1.8
@@ -14965,10 +14975,10 @@ snapshots:
dependencies:
zod: 4.5.4
dependencies:
transitivePeerDependencies:
- ai
- supports-color
@@ -14999,10 +15009,10 @@ snapshots:
zod: 4.4.3
optional: true
dependencies:
- '@ai-sdk/provider': 4.0.17
- '@ai-sdk/provider-utils': 5.0.43([email protected])
+ '@ai-sdk/provider': 4.0.22
+ '@ai-sdk/provider-utils': 5.0.54([email protected])
zod: 4.5.4
@@ -15022,9 +15032,9 @@ snapshots:
zod: 4.4.3
optional: true
dependencies:
- ai: 7.0.127([email protected])
+ ai: 7.0.128([email protected])
run: 2.1.2([email protected])
transitivePeerDependencies:
- typescript
@@ -15036,17 +15046,17 @@ snapshots:
zod: 4.4.3
optional: true
dependencies:
- '@ai-sdk/provider': 2.0.4
- '@ai-sdk/provider-utils': 3.0.37([email protected])
+ '@ai-sdk/provider': 2.0.5
+ '@ai-sdk/provider-utils': 3.0.41([email protected])
'@vercel/oidc': 3.1.0
zod: 3.25.76
dependencies:
- '@ai-sdk/provider': 2.0.4
- '@ai-sdk/provider-utils': 3.0.37([email protected])
+ '@ai-sdk/provider': 2.0.5
+ '@ai-sdk/provider-utils': 3.0.41([email protected])
'@vercel/oidc': 3.1.0
zod: 4.4.3
@@ -15057,26 +15067,26 @@ snapshots:
'@vercel/oidc': 3.2.0
zod: 4.6.5
dependencies:
- '@ai-sdk/provider': 4.0.21
- '@ai-sdk/provider-utils': 5.0.53([email protected])
+ '@ai-sdk/provider': 3.0.18
+ '@ai-sdk/provider-utils': 4.0.57([email protected])
'@vercel/oidc': 3.2.0
- zod: 4.5.4
+ zod: 4.6.5
dependencies:
- '@ai-sdk/provider': 4.0.21
- '@ai-sdk/provider-utils': 5.0.53([email protected])
+ '@ai-sdk/provider': 4.0.22
+ '@ai-sdk/provider-utils': 5.0.54([email protected])
'@vercel/oidc': 3.2.0
- zod: 4.6.5
+ zod: 4.5.4
dependencies:
- '@ai-sdk/provider': 4.0.17
- '@ai-sdk/provider-utils': 5.0.43([email protected])
+ '@ai-sdk/provider': 4.0.22
+ '@ai-sdk/provider-utils': 5.0.54([email protected])
'@vercel/oidc': 3.2.0
- zod: 4.5.4
+ zod: 4.6.5
dependencies:
@@ -15098,10 +15108,10 @@ snapshots:
zod: 4.4.3
optional: true
dependencies:
- '@ai-sdk/provider': 4.0.17
- '@ai-sdk/provider-utils': 5.0.43([email protected])
+ '@ai-sdk/provider': 4.0.22
+ '@ai-sdk/provider-utils': 5.0.54([email protected])
zod: 4.5.4
@@ -15111,10 +15121,10 @@ snapshots:
zod: 4.4.3
optional: true
dependencies:
- '@ai-sdk/provider': 4.0.17
- '@ai-sdk/provider-utils': 5.0.43([email protected])
+ '@ai-sdk/provider': 4.0.22
+ '@ai-sdk/provider-utils': 5.0.54([email protected])
cross-spawn: 7.0.6
pkce-challenge: 5.0.1
zod: 4.5.4
@@ -15140,23 +15150,23 @@ snapshots:
zod: 4.4.3
optional: true
dependencies:
- '@ai-sdk/provider': 4.0.17
- '@ai-sdk/provider-utils': 5.0.43([email protected])
+ '@ai-sdk/provider': 4.0.22
+ '@ai-sdk/provider-utils': 5.0.54([email protected])
zod: 4.5.4
dependencies:
- '@ai-sdk/provider': 4.0.17
- '@ai-sdk/provider-utils': 5.0.43([email protected])
+ '@ai-sdk/provider': 4.0.22
+ '@ai-sdk/provider-utils': 5.0.54([email protected])
zod: 4.6.5
dependencies:
- '@ai-sdk/provider': 4.0.17
+ '@ai-sdk/provider': 4.0.22
'@opentelemetry/api': 1.9.1
- ai: 7.0.105([email protected])
+ ai: 7.0.128([email protected])
transitivePeerDependencies:
- zod
@@ -15174,17 +15184,26 @@ snapshots:
secure-json-parse: 2.7.0
zod: 3.25.76
dependencies:
'@ai-sdk/provider': 2.0.4
'@standard-schema/spec': 1.1.0
eventsource-parser: 3.1.0
undici: 5.29.0
+ zod: 4.4.3
+ optional: true
+
+ dependencies:
+ '@ai-sdk/provider': 2.0.5
+ '@standard-schema/spec': 1.1.0
+ eventsource-parser: 3.1.0
+ undici: 5.29.0
zod: 3.25.76
dependencies:
- '@ai-sdk/provider': 2.0.4
+ '@ai-sdk/provider': 2.0.5
'@standard-schema/spec': 1.1.0
eventsource-parser: 3.1.0
undici: 5.29.0
@@ -15198,36 +15217,26 @@ snapshots:
undici: 6.28.1
zod: 4.6.5
dependencies:
- '@ai-sdk/provider': 4.0.17
+ '@ai-sdk/provider': 3.0.18
'@standard-schema/spec': 1.1.0
- '@workflow/serde': 4.1.0
eventsource-parser: 3.1.0
- undici: 7.30.0
- zod: 4.5.4
-
- dependencies:
- '@ai-sdk/provider': 4.0.17
- '@standard-schema/spec': 1.1.0
- '@workflow/serde': 4.1.0
- eventsource-parser: 3.1.0
- undici: 7.30.0
+ undici: 6.28.1
zod: 4.6.5
dependencies:
- '@ai-sdk/provider': 4.0.21
+ '@ai-sdk/provider': 4.0.22
'@standard-schema/spec': 1.1.0
'@workflow/serde': 4.1.0
eventsource-parser: 3.1.0
undici: 7.30.0
zod: 4.5.4
dependencies:
- '@ai-sdk/provider': 4.0.21
+ '@ai-sdk/provider': 4.0.22
'@standard-schema/spec': 1.1.0
'@workflow/serde': 4.1.0
eventsource-parser: 3.1.0
@@ -15241,16 +15250,21 @@ snapshots:
'@ai-sdk/[email protected]':
dependencies:
json-schema: 0.4.0
+ optional: true
+
+ '@ai-sdk/[email protected]':
+ dependencies:
+ json-schema: 0.4.0
'@ai-sdk/[email protected]':
dependencies:
json-schema: 0.4.0
- '@ai-sdk/[email protected]':
+ '@ai-sdk/[email protected]':
dependencies:
json-schema: 0.4.0
- '@ai-sdk/[email protected]':
+ '@ai-sdk/[email protected]':
dependencies:
json-schema: 0.4.0
@@ -15575,11 +15589,11 @@ snapshots:
'@balena/[email protected]': {}
dependencies:
- '@chat-adapter/state-redis': 4.41.0(@opentelemetry/[email protected])([email protected]([email protected]))([email protected])([email protected])
+ '@chat-adapter/state-redis': 4.41.0(@opentelemetry/[email protected])([email protected]([email protected]))([email protected])([email protected])
marked: 15.0.12
matrix-js-sdk: 41.9.0
node-html-parser: 7.1.0
@@ -15643,12 +15657,12 @@ snapshots:
dependencies:
- '@ai-sdk/provider': 2.0.4
+ '@ai-sdk/provider': 2.0.5
'@anthropic-ai/sdk': 0.39.0
'@browserbasehq/sdk': 2.16.0
'@modelcontextprotocol/sdk': 1.30.0(@cfworker/[email protected])([email protected])([email protected])
- ai: 5.0.259([email protected])
+ ai: 5.0.272([email protected])
devtools-protocol: 0.0.1642743
fetch-cookie: 3.2.0
@@ -15790,22 +15804,22 @@ snapshots:
'@changesets/types': 7.0.0
human-id: 4.2.1
dependencies:
'@googleapis/chat': 44.7.0([email protected])
'@googleapis/workspaceevents': 9.2.0([email protected])
transitivePeerDependencies:
- ai
- supports-color
- workflow
- zod
dependencies:
html-to-text: 9.0.5
jose: 6.2.3
mimetext: 3.0.28
@@ -15816,57 +15830,48 @@ snapshots:
- supports-color
- workflow
dependencies:
transitivePeerDependencies:
- ai
- supports-color
- workflow
- zod
- dependencies:
- transitivePeerDependencies:
- - ai
- - supports-color
- - zod
-
dependencies:
transitivePeerDependencies:
- ai
- supports-color
- zod
- optional: true
dependencies:
transitivePeerDependencies:
- ai
- supports-color
- zod
optional: true
dependencies:
transitivePeerDependencies:
- ai
- supports-color
- workflow
- zod
dependencies:
transitivePeerDependencies:
- ai
- bufferutil
@@ -15876,27 +15881,12 @@ snapshots:
- zod
optional: true
dependencies:
- transitivePeerDependencies:
- - ai
- - bufferutil
- - debug
- - supports-color
- - utf-8-validate
- - zod
- optional: true
-
- dependencies:
transitivePeerDependencies:
- ai
- bufferutil
@@ -15906,18 +15896,18 @@ snapshots:
- workflow
- zod
dependencies:
transitivePeerDependencies:
- ai
- supports-color
- workflow
- zod
dependencies:
redis: 5.12.1(@opentelemetry/[email protected])
transitivePeerDependencies:
- '@node-rs/xxhash'
@@ -15927,30 +15917,30 @@ snapshots:
- workflow
- zod
dependencies:
transitivePeerDependencies:
- ai
- supports-color
- workflow
- zod
dependencies:
transitivePeerDependencies:
- ai
- supports-color
- workflow
- zod
dependencies:
transitivePeerDependencies:
- ai
- supports-color
@@ -16400,11 +16390,11 @@ snapshots:
dependencies:
tslib: 2.8.1
dependencies:
transitivePeerDependencies:
- ai
- encoding
@@ -16422,26 +16412,26 @@ snapshots:
- react-native
- supports-color
dependencies:
eve: link:packages/eve
zod: 4.5.4
optionalDependencies:
transitivePeerDependencies:
- '@ai-sdk/workflow'
- '@workflow/ai'
- ai
- workflow
dependencies:
- ai: 7.0.127([email protected])
+ ai: 7.0.128([email protected])
octokit: 5.0.5
zod: 4.5.4
optionalDependencies:
eve: link:packages/eve
@@ -16713,11 +16703,11 @@ snapshots:
'@js-sdsl/[email protected]': {}
dependencies:
'@kapso/whatsapp-cloud-api': 0.2.3
transitivePeerDependencies:
- ai
- supports-color
@@ -16786,11 +16776,11 @@ snapshots:
'@langchain/[email protected]': {}
dependencies:
'@larksuiteoapi/node-sdk': 1.63.1([email protected])([email protected]([email protected]))
transitivePeerDependencies:
- ai
- bufferutil
@@ -16814,21 +16804,21 @@ snapshots:
- debug
- utf-8-validate
dependencies:
'@linqapp/sdk': 0.47.1
standardwebhooks: 1.1.1
'@linqapp/[email protected]':
dependencies:
standardwebhooks: 1.1.1
dependencies:
'@liveblocks/core': 3.23.0(@types/[email protected])
'@liveblocks/node': 3.23.0(@types/[email protected])
transitivePeerDependencies:
- '@types/json-schema'
- encoding
@@ -17132,10 +17122,10 @@ snapshots:
'@nolyfill/[email protected]': {}
dependencies:
optionalDependencies:
react: 19.2.6
transitivePeerDependencies:
@@ -17207,73 +17197,7 @@ snapshots:
pkg-types: 2.3.3
semver: 7.8.5
- dependencies:
- '@nuxt/devtools-wizard': 3.4.2
- '@vue/devtools-core': 8.2.1([email protected]([email protected]))
- '@vue/devtools-kit': 8.2.1
- birpc: 4.2.0
- consola: 3.4.2
- destr: 2.0.5
- error-stack-parser-es: 2.0.1
- execa: 8.0.1
- fast-npm-meta: 2.2.0
- get-port-please: 3.2.0
- hookable: 6.1.1
- image-meta: 0.2.2
- is-installed-globally: 1.0.0
- launch-editor: 2.14.1
- local-pkg: 1.2.1
- magicast: 0.5.5
- nypm: 0.6.10
- ohash: 2.0.12
- pathe: 2.0.3
- perfect-debounce: 2.1.0
- pkg-types: 2.3.3
- semver: 7.8.5
- simple-git: 3.36.0([email protected])
- sirv: 3.0.2
- structured-clone-es: 2.0.1
- tinyglobby: 0.2.17
- which: 6.0.1
- ws: 8.21.3([email protected])
- transitivePeerDependencies:
- - '@azure/app-configuration'
- - '@azure/cosmos'
- - '@azure/data-tables'
- - '@azure/identity'
- - '@azure/keyvault-secrets'
- - '@azure/storage-blob'
- - '@capacitor/preferences'
- - '@deno/kv'
- - '@netlify/blobs'
- - '@planetscale/database'
- - '@upstash/redis'
- - '@vercel/blob'
- - '@vercel/functions'
- - '@vercel/kv'
- - aws4fetch
- - bufferutil
- - db0
- - idb-keyval
- - ioredis
- - magic-string
- - oxc-parser
- - rolldown
- - supports-color
- - unplugin
- - uploadthing
- - utf-8-validate
- - vue
- optional: true
-
dependencies:
'@nuxt/devtools-wizard': 3.4.2
@@ -17368,7 +17292,7 @@ snapshots:
- rolldown
- unplugin
- '@nuxt/[email protected](373e6c36b47f163c0b84a471567193af)':
+ '@nuxt/[email protected](2467243e5e4e5bbd99c0d73c82272e45)':
dependencies:
'@nuxt/devalue': 2.0.2
@@ -17385,9 +17309,9 @@ snapshots:
klona: 2.0.6
mocked-exports: 0.1.1
nostics: 1.2.0
- nuxt: 4.5.2(24c7457ca6c06a442e66ba8ec8c3a0d4)
+ nuxt: 4.5.2(061f1cf7a0f80874150f1bf0055f39b3)
nypm: 0.6.10
ohash: 2.0.12
pathe: 2.0.3
@@ -17453,8 +17377,9 @@ snapshots:
- vite
- webpack
- xml2js
+ optional: true
- '@nuxt/[email protected](b4006145566056bafcc84d4debed1413)':
+ '@nuxt/[email protected](373e6c36b47f163c0b84a471567193af)':
dependencies:
'@nuxt/devalue': 2.0.2
@@ -17471,9 +17396,9 @@ snapshots:
klona: 2.0.6
mocked-exports: 0.1.1
nostics: 1.2.0
+ nuxt: 4.5.2(24c7457ca6c06a442e66ba8ec8c3a0d4)
nypm: 0.6.10
ohash: 2.0.12
pathe: 2.0.3
@@ -17481,7 +17406,7 @@ snapshots:
std-env: 4.2.0
ufo: 1.6.4
vue: 3.5.42([email protected])
vue-bundle-renderer: 2.4.0
vue-devtools-stub: 0.1.0
@@ -17539,7 +17464,6 @@ snapshots:
- vite
- webpack
- xml2js
- optional: true
dependencies:
@@ -17558,7 +17482,7 @@ snapshots:
rc9: 3.1.0
std-env: 4.2.0
dependencies:
@@ -17576,7 +17500,7 @@ snapshots:
knitwork: 1.3.0
mlly: 1.8.2
mocked-exports: 0.1.1
- nuxt: 4.5.2(24c7457ca6c06a442e66ba8ec8c3a0d4)
+ nuxt: 4.5.2(061f1cf7a0f80874150f1bf0055f39b3)
nypm: 0.6.10
pathe: 2.0.3
pkg-types: 2.3.3
@@ -17626,8 +17550,9 @@ snapshots:
- vue-tsc
- webpack
- yaml
+ optional: true
- '@nuxt/[email protected](f3dac387c419865d25be8b8aaf8cebc8)':
dependencies:
@@ -17645,7 +17570,7 @@ snapshots:
knitwork: 1.3.0
mlly: 1.8.2
mocked-exports: 0.1.1
+ nuxt: 4.5.2(24c7457ca6c06a442e66ba8ec8c3a0d4)
nypm: 0.6.10
pathe: 2.0.3
pkg-types: 2.3.3
@@ -17695,7 +17620,6 @@ snapshots:
- vue-tsc
- webpack
- yaml
- optional: true
'@octokit/[email protected]':
dependencies:
@@ -17862,9 +17786,9 @@ snapshots:
'@octokit/request-error': 7.1.2
'@octokit/webhooks-methods': 6.0.0
dependencies:
eve: link:packages/eve
zod: 4.4.3
transitivePeerDependencies:
@@ -18335,12 +18259,12 @@ snapshots:
nice-grpc: 2.1.17
nice-grpc-common: 2.0.4
dependencies:
'@spectrum-ts/core': 10.0.0([email protected])([email protected])
mime-types: 3.0.2
transitivePeerDependencies:
- ai
@@ -19656,10 +19580,10 @@ snapshots:
'@repeaterjs/[email protected]': {}
dependencies:
hast-util-to-html: 9.0.5
mdast-util-to-hast: 13.2.1
@@ -20945,16 +20869,16 @@ snapshots:
d3-selection: 3.0.0
d3-transition: 3.0.1([email protected])
dependencies:
'@upstash/agentkit-sdk': 0.9.0(@upstash/[email protected])
'@upstash/redis': 1.38.4
- ai: 7.0.127([email protected])
+ ai: 7.0.128([email protected])
zod: 4.5.4
dependencies:
- '@upstash/agentkit-ai-sdk': 0.9.0([email protected]([email protected]))
+ '@upstash/agentkit-ai-sdk': 0.9.0([email protected]([email protected]))
'@upstash/agentkit-sdk': 0.9.0(@upstash/[email protected])
'@upstash/redis': 1.38.4
eve: link:packages/eve
@@ -20985,10 +20909,10 @@ snapshots:
dependencies:
eve: link:packages/eve
dependencies:
hast-util-from-html: 2.0.3
hast-util-to-html: 9.0.5
hast-util-to-mdast: 10.1.2
@@ -21004,7 +20928,7 @@ snapshots:
dependencies:
'@ai-sdk/anthropic': 1.2.12([email protected])
'@vercel/sandbox': 1.10.2
- ai: 5.0.259([email protected])
+ ai: 5.0.272([email protected])
chalk: 5.6.2
commander: 12.1.0
dockerode: 4.0.12([email protected])
@@ -21029,11 +20953,11 @@ snapshots:
h3: 1.15.11
- '@vercel/[email protected](e9f01ee4848ee8142aacdb51dfc5a6a4)':
optionalDependencies:
+ nuxt: 4.5.2(061f1cf7a0f80874150f1bf0055f39b3)
react: 19.2.6
svelte: 5.57.0(@typescript-eslint/[email protected])
vue: 3.5.42([email protected])
@@ -21083,57 +21007,47 @@ snapshots:
dependencies:
execa: 5.1.1
dependencies:
'@vercel/oidc': 3.8.1
optionalDependencies:
- '@ai-sdk/mcp': 2.0.52([email protected])
- '@auth/core': 0.41.2
- ai: 7.0.127([email protected])
- eve: link:packages/eve
-
- dependencies:
- '@vercel/oidc': 3.8.8
- optionalDependencies:
- '@ai-sdk/mcp': 2.0.52([email protected])
+ '@ai-sdk/mcp': 2.0.67([email protected])
'@auth/core': 0.41.2
- ai: 7.0.105([email protected])
+ ai: 7.0.128([email protected])
eve: link:packages/eve
dependencies:
'@vercel/oidc': 3.8.8
optionalDependencies:
- '@ai-sdk/mcp': 2.0.52([email protected])
+ '@ai-sdk/mcp': 2.0.67([email protected])
'@auth/core': 0.41.2
- ai: 7.0.127([email protected])
+ ai: 7.0.128([email protected])
eve: link:packages/eve
dependencies:
'@vercel/oidc': 3.8.9
optionalDependencies:
- '@ai-sdk/mcp': 2.0.52([email protected])
+ '@ai-sdk/mcp': 2.0.67([email protected])
'@auth/core': 0.41.2
'@modelcontextprotocol/sdk': 1.30.0(@cfworker/[email protected])([email protected])([email protected])
- ai: 7.0.105([email protected])
+ ai: 7.0.128([email protected])
eve: link:packages/eve
dependencies:
'@vercel/oidc': 3.8.9
optionalDependencies:
- '@ai-sdk/mcp': 2.0.52([email protected])
+ '@ai-sdk/mcp': 2.0.67([email protected])
'@auth/core': 0.41.2
'@modelcontextprotocol/sdk': 1.30.0(@cfworker/[email protected])([email protected])([email protected])
- ai: 7.0.127([email protected])
+ ai: 7.0.128([email protected])
'@vercel/[email protected]': {}
@@ -21160,7 +21074,7 @@ snapshots:
'@types/react': 19.2.15
'@vercel/oidc': 3.8.0
- ai: 6.0.285([email protected])
+ ai: 6.0.301([email protected])
class-variance-authority: 0.7.1
clsx: 2.1.1
commander: 14.0.3
@@ -21324,11 +21238,11 @@ snapshots:
dependencies:
zod: 4.5.4
- '@vercel/[email protected](e9f01ee4848ee8142aacdb51dfc5a6a4)':
optionalDependencies:
+ nuxt: 4.5.2(061f1cf7a0f80874150f1bf0055f39b3)
react: 19.2.6
svelte: 5.57.0(@typescript-eslint/[email protected])
vue: 3.5.42([email protected])
@@ -21656,10 +21570,10 @@ snapshots:
ulid: 3.0.2
zod: 4.5.4
dependencies:
transitivePeerDependencies:
- ai
- supports-color
@@ -21712,19 +21626,19 @@ snapshots:
dependencies:
humanize-ms: 1.2.1
dependencies:
- '@ai-sdk/gateway': 2.0.153([email protected])
- '@ai-sdk/provider': 2.0.4
- '@ai-sdk/provider-utils': 3.0.37([email protected])
+ '@ai-sdk/gateway': 2.0.163([email protected])
+ '@ai-sdk/provider': 2.0.5
+ '@ai-sdk/provider-utils': 3.0.41([email protected])
'@opentelemetry/api': 1.9.0
zod: 3.25.76
dependencies:
- '@ai-sdk/gateway': 2.0.153([email protected])
- '@ai-sdk/provider': 2.0.4
- '@ai-sdk/provider-utils': 3.0.37([email protected])
+ '@ai-sdk/gateway': 2.0.163([email protected])
+ '@ai-sdk/provider': 2.0.5
+ '@ai-sdk/provider-utils': 3.0.41([email protected])
'@opentelemetry/api': 1.9.0
zod: 4.4.3
@@ -21736,25 +21650,26 @@ snapshots:
'@opentelemetry/api': 1.9.1
zod: 4.6.5
dependencies:
- '@ai-sdk/gateway': 4.0.85([email protected])
- '@ai-sdk/provider': 4.0.17
- '@ai-sdk/provider-utils': 5.0.43([email protected])
- zod: 4.5.4
+ '@ai-sdk/gateway': 3.0.210([email protected])
+ '@ai-sdk/provider': 3.0.18
+ '@ai-sdk/provider-utils': 4.0.57([email protected])
+ '@opentelemetry/api': 1.9.1
+ zod: 4.6.5
dependencies:
- '@ai-sdk/gateway': 4.0.103([email protected])
- '@ai-sdk/provider': 4.0.21
- '@ai-sdk/provider-utils': 5.0.53([email protected])
+ '@ai-sdk/gateway': 4.0.104([email protected])
+ '@ai-sdk/provider': 4.0.22
+ '@ai-sdk/provider-utils': 5.0.54([email protected])
zod: 4.5.4
dependencies:
- '@ai-sdk/gateway': 4.0.103([email protected])
- '@ai-sdk/provider': 4.0.21
- '@ai-sdk/provider-utils': 5.0.53([email protected])
+ '@ai-sdk/gateway': 4.0.104([email protected])
+ '@ai-sdk/provider': 4.0.22
+ '@ai-sdk/provider-utils': 5.0.54([email protected])
zod: 4.6.5
@@ -22295,12 +22210,12 @@ snapshots:
dependencies:
sendblue: 3.16.1
dependencies:
'@workflow/serde': 4.1.0-beta.2
mdast-util-to-string: 4.0.0
@@ -22310,12 +22225,12 @@ snapshots:
remend: 1.3.1
unified: 11.0.5
optionalDependencies:
- ai: 7.0.127([email protected])
+ ai: 7.0.128([email protected])
zod: 4.5.4
transitivePeerDependencies:
- supports-color
dependencies:
'@workflow/serde': 4.1.0-beta.2
mdast-util-to-string: 4.0.0
@@ -22325,13 +22240,13 @@ snapshots:
remend: 1.3.1
unified: 11.0.5
optionalDependencies:
- ai: 7.0.105([email protected])
+ ai: 7.0.128([email protected])
zod: 4.5.4
transitivePeerDependencies:
- supports-color
optional: true
dependencies:
'@workflow/serde': 4.1.0-beta.2
mdast-util-to-string: 4.0.0
@@ -22341,23 +22256,7 @@ snapshots:
remend: 1.3.1
unified: 11.0.5
optionalDependencies:
- ai: 7.0.127([email protected])
- zod: 4.5.4
- transitivePeerDependencies:
- - supports-color
- optional: true
-
- dependencies:
- '@workflow/serde': 4.1.0-beta.2
- mdast-util-to-string: 4.0.0
- remark-gfm: 4.0.1([email protected])
- remark-parse: 11.0.0([email protected])
- remark-stringify: 11.0.0
- remend: 1.3.1
- unified: 11.0.5
- optionalDependencies:
- ai: 7.0.127([email protected])
+ ai: 7.0.128([email protected])
zod: 4.5.4
transitivePeerDependencies:
- supports-color
@@ -23790,20 +23689,20 @@ snapshots:
dependencies:
- ai: 7.0.127([email protected])
+ ai: 7.0.128([email protected])
eve: link:packages/eve
dependencies:
- ai: 7.0.127([email protected])
+ ai: 7.0.128([email protected])
nitro: 3.0.260903-beta
undici: 8.9.0
optionalDependencies:
'@opentelemetry/api': 1.9.1
dd-trace: 6.13.0
just-bash: 3.1.0([email protected])
microsandbox: 0.5.5
@@ -26637,7 +26536,7 @@ snapshots:
unenv: 2.0.0-rc.24
unstorage: 2.0.0-alpha.10
dependencies:
'@cloudflare/kv-asset-handler': 0.4.2
'@rollup/plugin-alias': 6.0.0([email protected])
@@ -26933,16 +26832,16 @@ snapshots:
- [email protected](24c7457ca6c06a442e66ba8ec8c3a0d4):
+ [email protected](061f1cf7a0f80874150f1bf0055f39b3):
dependencies:
- '@nuxt/nitro-server': 4.5.2(373e6c36b47f163c0b84a471567193af)
+ '@nuxt/nitro-server': 4.5.2(2467243e5e4e5bbd99c0d73c82272e45)
'@nuxt/schema': 4.5.2
'@vue/shared': 3.5.42
chokidar: 5.0.0
@@ -27073,17 +26972,18 @@ snapshots:
- webpack
- xml2js
- yaml
+ optional: true
+ [email protected](24c7457ca6c06a442e66ba8ec8c3a0d4):
dependencies:
- '@nuxt/nitro-server': 4.5.2(b4006145566056bafcc84d4debed1413)
+ '@nuxt/nitro-server': 4.5.2(373e6c36b47f163c0b84a471567193af)
'@nuxt/schema': 4.5.2
- '@nuxt/vite-builder': 4.5.2(f3dac387c419865d25be8b8aaf8cebc8)
'@vue/shared': 3.5.42
chokidar: 5.0.0
@@ -27214,7 +27114,6 @@ snapshots:
- webpack
- xml2js
- yaml
- optional: true
dependencies:
@@ -27312,8 +27211,8 @@ snapshots:
dependencies:
- '@ai-sdk/provider': 2.0.4
- '@ai-sdk/provider-utils': 3.0.37([email protected])
+ '@ai-sdk/provider': 2.0.5
+ '@ai-sdk/provider-utils': 3.0.41([email protected])
zod: 4.4.3
optional: true
@@ -30054,25 +29953,6 @@ snapshots:
'@unrs/resolver-binding-win32-ia32-msvc': 1.12.2
'@unrs/resolver-binding-win32-x64-msvc': 1.12.2
- dependencies:
- anymatch: 3.1.3
- chokidar: 5.0.0
- destr: 2.0.5
- h3: 1.15.11
- lru-cache: 11.5.2
- node-fetch-native: 1.6.7
- ofetch: 1.5.1
- ufo: 1.6.4
- optionalDependencies:
- '@upstash/redis': 1.38.4
- '@vercel/blob': 2.8.0
- '@vercel/functions': 3.9.8(@aws-sdk/[email protected])([email protected])
- aws4fetch: 1.0.20
- db0: 0.4.1
- ioredis: 5.11.1([email protected])
- optional: true
-
dependencies:
anymatch: 3.1.3
diff --git a/pnpm-workspace.yaml b/pnpm-workspace.yaml
index 7dabd2ac9..dcd1585d4 100644
--- a/pnpm-workspace.yaml
+++ b/pnpm-workspace.yaml
@@ -29,14 +29,14 @@ ignoredOptionalDependencies:
- node-liblzma
catalog:
- "@ai-sdk/anthropic": "^4.0.49"
- "@ai-sdk/code-mode": "^1.0.62"
- "@ai-sdk/google": "^4.0.64"
- "@ai-sdk/mcp": "^2.0.45"
- "@ai-sdk/openai": "^4.0.60"
- "@ai-sdk/otel": "^1.0.93"
- "@ai-sdk/provider": "^4.0.10"
- "@ai-sdk/provider-utils": "^5.0.36"
+ "@ai-sdk/anthropic": "^4.0.72"
+ "@ai-sdk/code-mode": "^1.0.85"
+ "@ai-sdk/google": "^4.0.88"
+ "@ai-sdk/mcp": "^2.0.67"
+ "@ai-sdk/openai": "^4.0.84"
+ "@ai-sdk/otel": "^1.0.128"
+ "@ai-sdk/provider": "^4.0.22"
+ "@ai-sdk/provider-utils": "^5.0.54"
"@opentelemetry/context-async-hooks": "2.6.1"
"@opentelemetry/core": "2.6.1"
"@opentelemetry/sdk-trace-base": "2.6.1"
@@ -47,7 +47,7 @@ catalog:
"@vercel/otel": "2.1.3"
"@vercel/sandbox": "3.3.0"
"@workflow/world-postgres": "5.0.1"
- ai: "^7.0.116"
+ ai: "^7.0.128"
better-auth: "1.6.26"
marked: "17.0.6"
next: "16.3.4"
diff --git a/research/automatic-tool-approval.md b/research/automatic-tool-approval.md
index 4926cc8e8..334df9ec1 100644
--- a/research/automatic-tool-approval.md
+++ b/research/automatic-tool-approval.md
@@ -1,14 +1,14 @@
---
issue: "TBD (maintainer-requested implementation; no matching issue found)"
status: implemented
-last_updated: "2026-09-17"
+last_updated: "2026-10-06"
---
# Automatic tool approval
## Authoring API
-Add `auto({ model?, instructions?, criteria? })` to `eve/tools/approval`. `model` accepts any AI SDK `Experimental_EvaluationModel`, including a string resolved through the configured default provider, and defaults to `typesafe-ai/jev`. `instructions` and `criteria.clear` / `criteria.caution` override the classifier text.
+Add `auto({ model?, instructions?, criteria? })` to `eve/tools/approval`. `model` accepts any AI SDK `Experimental_DecisionModel`, including a string resolved through the configured default provider, and defaults to `typesafe-ai/jev`. `instructions` and `criteria.clear` / `criteria.caution` override the classifier text.
```ts
import { defineTool } from "eve/tools";
@@ -20,14 +20,14 @@ export default defineTool({
});
```
-The helper asks one constrained choice question with `clear` and `caution` outcomes. `clear` returns `"approved"`; `caution` returns `"user-approval"`. Provider errors, invalid answers, timeouts, oversized input, and non-serializable tool input fail closed to user approval. Cancelling the active turn cancels evaluation instead of producing a prompt.
+The helper asks one constrained choice question with `clear` and `caution` outcomes. `clear` returns `"approved"`; `caution` returns `"user-approval"`. Provider errors, invalid answers, timeouts, oversized input, and non-serializable tool input fail closed to user approval. Cancelling the active turn cancels decision instead of producing a prompt.
## Evidence and cancellation
-The evaluation state contains the exact tool name and input. Assistant prose and conversation history are excluded. `ApprovalContext` adds `abortSignal` so the helper and custom asynchronous approval policies stop with the active turn.
+The decision state contains the exact tool name and input. Assistant prose and conversation history are excluded. `ApprovalContext` adds `abortSignal` so the helper and custom asynchronous approval policies stop with the active turn.
## Policy
The default classifier returns caution for dangerous or unclear effects, including destructive data loss, credential exposure, financial transactions, deployments, public changes, external communication, privilege or system changes, and concealed execution.
-`auto()` is not authorization. Applications must still enforce identity, tenancy, service permissions, allowlists, and other deterministic requirements. Tool input is sent to the selected evaluation provider, so authors must not use it for inputs the provider is not permitted to receive.
+`auto()` is not authorization. Applications must still enforce identity, tenancy, service permissions, allowlists, and other deterministic requirements. Tool input is sent to the selected decision provider, so authors must not use it for inputs the provider is not permitted to receive.
diff --git a/research/evaluation-model-judges.md b/research/decision-model-judges.md
similarity index 83%
rename from research/evaluation-model-judges.md
rename to research/decision-model-judges.md
index 8dca06151..95f16643a 100644
--- a/research/evaluation-model-judges.md
+++ b/research/decision-model-judges.md
@@ -1,12 +1,12 @@
---
issue: "TBD (maintainer-requested research; no matching issue found)"
status: implemented
-last_updated: "2026-09-18"
+last_updated: "2026-10-06"
---
-# Evaluation models for eval judges
+# Decision models for eval judges
-`t.judge(...)` uses the existing `evaluate` wrapper from `eve/ai`. A criteria
+`t.judge(...)` uses the existing `decide` wrapper from `eve/ai`. A criteria
string becomes a boolean question; an explicit question returns one assertion
handle; `{ state?, questions }` returns named handles backed by one request.
All calls start immediately and retain existing soft scoring, labels, thresholds,
@@ -20,8 +20,8 @@ score selected-option equality as 0 or 1; the expectation is never sent to the
model. Batches share one state, preserve question order, and fail together if
SDK validation or the provider fails.
-`judge.model` accepts an evaluation model ID or instance and defaults to the
-shared evaluator, currently `typesafe-ai/jev`. Per-call settings override the
+`judge.model` accepts a decision model ID or instance and defaults to the
+shared decider, currently `typesafe-ai/jev`. Per-call settings override the
resolved eval/project configuration. Provider options retain their existing
`modelOptions.providerOptions` shape. AI SDK owns adapters, authentication,
validation, and retries. The eval signal cancels provider I/O and bounds how long
@@ -29,7 +29,7 @@ the runner waits during assertion finalization.
The implementation reuses the assertion collector, diagnostics, verdicts, and
reporters. Metadata records raw answers and normalized scores; batch token usage
-is explicitly shared. Evaluation does not promise prose rationales.
+is explicitly shared. Decision does not promise prose rationales.
Remove autoevals, its adapter, and its public namespace. Migrate criteria judges
to the callable form and use explicit state/questions for specialized rubrics.
diff --git a/research/jev-decision-models.md b/research/jev-decision-models.md
index 6298d1fa9..4a0d32e81 100644
--- a/research/jev-decision-models.md
+++ b/research/jev-decision-models.md
@@ -1,31 +1,31 @@
---
issue: "TBD (maintainer-requested research; no matching issue found)"
status: implemented
-last_updated: "2026-09-17"
+last_updated: "2026-10-06"
---
-# Evaluation model routing in eve
+# Decision model routing in eve
## Recommendation
Expose automatic model selection as `auto` from `eve/models` and standalone
-`evaluate` from `eve/ai`. Build both on AI SDK's `experimental_evaluate` API.
-`evaluate` accepts the SDK's state, typed questions, and request options, with an
+`decide` from `eve/ai`. Build both on AI SDK's `experimental_decide` API.
+`decide` accepts the SDK's state, typed questions, and request options, with an
optional model defaulting to `typesafe-ai/jev`. It returns the SDK's typed answers
and response metadata.
-`auto` calls the shared `evaluate` wrapper. Model strings use the configured AI
+`auto` calls the shared `decide` wrapper. Model strings use the configured AI
SDK default provider; without an override, the wrapper resolves eve's local
Gateway connection when available and otherwise leaves Gateway resolution to the
SDK. Explicit provider instances retain their own authentication.
-Tool authors can call `evaluate` with structured state and pass `ctx.abortSignal`.
+Tool authors can call `decide` with structured state and pass `ctx.abortSignal`.
Application code can use it without an active eve session. Standalone calls are
not cached; durable per-turn selection remains specific to `auto`.
-AI SDK's evaluation model specification remains experimental and can change in
+AI SDK's decision model specification remains experimental and can change in
patch releases. The implemented authoring API is documented in [Automatic Model
-Selection](../docs/guides/evaluate.md).
+Selection](../docs/guides/decide.md).
```ts
import { defineAgent } from "eve";
@@ -43,14 +43,14 @@ export default defineAgent({
## Why AI SDK is the boundary
-AI SDK 7.0.105 adds `experimental_evaluate` and the v4 experimental evaluation
-model contract. The core function accepts a string or
-`Experimental_EvaluationModel`, validates input and output, checks supported
+AI SDK 7.0.128 provides `experimental_decide` (shipped as `experimental_evaluate`
+in 7.0.105 and since renamed) and the v4 experimental decision model contract. The core function accepts a string or
+`Experimental_DecisionModel`, validates input and output, checks supported
question types, retries transient provider failures, propagates cancellation,
and normalizes usage and response metadata.
-[AI SDK evaluation](https://ai-sdk.dev/docs/ai-sdk-core/evaluation).
+[AI SDK decision](https://ai-sdk.dev/docs/ai-sdk-core/decisions).
-This gives evaluation models the same selection rules as other AI SDK models:
+This gives decision models the same selection rules as other AI SDK models:
- A string resolves through `globalThis.AI_SDK_DEFAULT_PROVIDER`. With no
override, AI SDK uses Vercel AI Gateway. Gateway credentials and OIDC remain
@@ -69,7 +69,7 @@ provider metadata parsing from eve.
```mermaid
flowchart LR
P[Incoming prompt] --> A[auto]
- A --> E[AI SDK evaluate]
+ A --> E[AI SDK decide]
E -->|string ID| G[Default provider / Gateway]
E -->|model instance| D[Installed provider]
G --> C[Choice answer]
@@ -81,7 +81,7 @@ flowchart LR
## What Jev provides
TypeSafe describes Jev as a System One decision model trained with Reinforcement
-Learning for Calibrated Decisions. It evaluates structured state against
+Learning for Calibrated Decisions. It decides structured state against
constrained questions instead of generating prose. The native TypeSafe API
describes three primitives:
@@ -97,7 +97,7 @@ Sources: [System One](https://docs.typesafe.ai/concepts/system-one),
[Score](https://docs.typesafe.ai/primitives/score), and
[Noul](https://docs.typesafe.ai/primitives/noul).
-Every question in one request sees the same state and is evaluated independently
+Every question in one request sees the same state and is decided independently
by native Jev. Question IDs correlate inputs and outputs but are not model
instructions. A question that targets one item in structured state must identify
that item in its instructions. Related questions can share one request; unrelated
@@ -112,16 +112,16 @@ router must therefore make no confidence-threshold promise across providers.
Constrained output prevents fabricated option names after SDK validation, but it
does not guarantee the selected option is correct. Model descriptions and labeled
-evaluation data remain part of the routing policy. Calibration describes behavior
+decision data remain part of the routing policy. Calibration describes behavior
across a dataset rather than certainty for one answer.
[Confidence](https://docs.typesafe.ai/confidence).
## Provider behavior
-AI SDK documents native TypeSafe evaluation through
-`typeSafeAi.evaluationModel("jev-latest")`. OpenAI, Anthropic, and Google expose
-evaluation model adapters that use structured language-model output. Those
-adapters evaluate all questions in one prompt, do not reproduce Jev's native
+AI SDK documents native TypeSafe decision through
+`typeSafeAi.decisionModel("jev-latest")`. OpenAI, Anthropic, and Google expose
+decision model adapters that use structured language-model output. Those
+adapters decide all questions in one prompt, do not reproduce Jev's native
independent-question execution, and omit Choice and Score distributions.
The [Gateway model catalog](https://vercel.com/ai-gateway/models/jev) uses
@@ -153,7 +153,7 @@ Provider model objects stay in authored configuration and are resolved again fro
the key after resume. New turns and child sessions make independent choices.
The active execution `AbortSignal` is passed through dynamic model context into
-AI SDK evaluation. This prevents a cancelled turn from completing routing or
+AI SDK decision. This prevents a cancelled turn from completing routing or
persisting a late choice.
Per-option `reasoning` is part of the selected dynamic model result, so the
@@ -163,12 +163,12 @@ direct provider behavior.
## Scope boundaries
-eve owns the `evaluate` wrapper and its default model and authentication behavior.
-AI SDK owns evaluation schemas, validation, retries, errors, and result metadata.
+eve owns the `decide` wrapper and its default model and authentication behavior.
+AI SDK owns decision schemas, validation, retries, errors, and result metadata.
The integration adds no TypeSafe-specific `decide` function, direct HTTP client,
provider fallback, confidence threshold, or built-in agent-callable decision tool.
No live Jev inference or independent quality, latency, or cost benchmark was run
-for this research. Deterministic tests use AI SDK evaluation model mocks and
+for this research. Deterministic tests use AI SDK decision model mocks and
exercise model selection, validation, cancellation, retention, provider model
instances, reasoning, and packaging.
diff --git a/research/subagent-tool-projection.md b/research/subagent-tool-projection.md
index 92e6eb1d6..eafa524b4 100644
--- a/research/subagent-tool-projection.md
+++ b/research/subagent-tool-projection.md
@@ -64,4 +64,4 @@ return ctx.agent(target, { message: task });
The owner snapshots metadata when it starts the workflow run. Older in-flight workflow payloads default to an empty metadata registry. `ctx.agent()` still validates availability at invocation time, so a dynamic agent that becomes unavailable after the snapshot cannot be invoked through stale metadata.
-The provided `agentRouter()` workflow tool filters `ctx.agents` to entries with non-empty descriptions, sends its input message and that description map to the default evaluation model (`typesafe-ai/jev`), then invokes the selected name. It forwards an optional output schema and rejects an empty target map before evaluation. Authors use `defineWorkflowTool` directly when routing requires a subset, custom instructions, or a non-default evaluator.
+The provided `agentRouter()` workflow tool filters `ctx.agents` to entries with non-empty descriptions, sends its input message and that description map to the default decision model (`typesafe-ai/jev`), then invokes the selected name. It forwards an optional output schema and rejects an empty target map before evaluation. Authors use `defineWorkflowTool` directly when routing requires a subset, custom instructions, or a non-default evaluator.