Skip to content

A local-first multi-agent app development platform, built on PenguinHarness

fix(core,cli): AgentHub 0.4.9 so GPT-5.6 reads a large screenshot instead of rejecting it (#580)

Yaowei Zhenga64d08a25803
diff --git a/changelog/unreleased/2026-09-02-agenthub-0.4.9.md b/changelog/unreleased/2026-09-02-agenthub-0.4.9.md
new file mode 100644
index 00000000..5c7b7945
--- /dev/null
+++ b/changelog/unreleased/2026-09-02-agenthub-0.4.9.md
@@ -0,0 +1,16 @@
+# AgentHub 0.4.9: GPT-5.6 reads a large screenshot instead of rejecting it
+
+- **Date:** 2026-09-02
+- **Type:** fix
+- **Scope:** `core`, `cli`
+- **PR:** [#580](https://github.com/Prism-Shadow/penguin-harness/pull/580)
+- **Issue:** [#573](https://github.com/Prism-Shadow/penguin-harness/issues/573)
+
+[中文版](2026-09-02-agenthub-0.4.9.zh.md)
+
+`@prismshadow/agenthub` moved from 0.4.8 to 0.4.9 (agenthub [#194](https://github.com/Prism-Shadow/agenthub/pull/194)). Attaching a large screenshot to a `gpt-5.6-*` session no longer fails with `400 The image you provided requires N patches after processing, exceeding the limit of 30000`: the client reads the image's dimensions from the data URL's header and sends it at `detail: "high"` when it would exceed the limit at the model's default detail, so the API fits it into 2,500 patches and reads it. An image within the limit goes out as before, and no other model is affected.
+
+## Details
+
+- The fix lives in AgentHub's GPT-5.6, OpenAI Responses and OpenAI Chat clients, so the official `gpt-5.6-*` rows and the OpenRouter `openai/gpt-5.6-*` rows are covered alike, for an image in a prompt and in a tool result.
+- `pnpm-workspace.yaml`'s `minimumReleaseAgeExclude` entry names 0.4.9 in place of 0.4.8. A bump always lands inside pnpm's release-age gate, so the entry moves with each bump rather than being removed as its earlier note said.
diff --git a/changelog/unreleased/2026-09-02-agenthub-0.4.9.zh.md b/changelog/unreleased/2026-09-02-agenthub-0.4.9.zh.md
new file mode 100644
index 00000000..2366000e
--- /dev/null
+++ b/changelog/unreleased/2026-09-02-agenthub-0.4.9.zh.md
@@ -0,0 +1,16 @@
+# AgentHub 0.4.9:GPT-5.6 能读取大尺寸截图,不再拒收
+
+- **Date:** 2026-09-02
+- **Type:** fix
+- **Scope:** `core`, `cli`
+- **PR:** [#580](https://github.com/Prism-Shadow/penguin-harness/pull/580)
+- **Issue:** [#573](https://github.com/Prism-Shadow/penguin-harness/issues/573)
+
+[English](2026-09-02-agenthub-0.4.9.md)
+
+`@prismshadow/agenthub` 从 0.4.8 升到 0.4.9(agenthub [#194](https://github.com/Prism-Shadow/agenthub/pull/194))。在 `gpt-5.6-*` 会话里附上大尺寸截图不再报 `400 The image you provided requires N patches after processing, exceeding the limit of 30000`:client 从 data URL 的文件头读出图片尺寸,若按模型默认细节会超限,就以 `detail: "high"` 发送,API 便把它缩到 2,500 patch 之内读取。未超限的图片照旧发送,其他模型不受影响。
+
+## 细节
+
+- 修复位于 AgentHub 的 GPT-5.6、OpenAI Responses 与 OpenAI Chat 三个 client,因此官方 `gpt-5.6-*` 行与 OpenRouter 的 `openai/gpt-5.6-*` 行同样覆盖,提示词与工具返回里的图片一并适用。
+- `pnpm-workspace.yaml` 的 `minimumReleaseAgeExclude` 条目由 0.4.8 改为 0.4.9。每次 bump 都落在 pnpm 的发布时限之内,所以这条随 bump 移动,而不是像原注释说的那样删除。
diff --git a/packages/cli/package.json b/packages/cli/package.json
index 9fe6aa21..b55e606b 100644
--- a/packages/cli/package.json
+++ b/packages/cli/package.json
@@ -36,7 +36,7 @@
"@penguinharness/use-bento-slides": "workspace:*",
"@penguinharness/use-firecrawl": "workspace:*",
"@penguinharness/use-claude-code": "workspace:*",
- "@prismshadow/agenthub": "^0.4.8",
+ "@prismshadow/agenthub": "^0.4.9",
"@prismshadow/penguin-core": "workspace:*",
"@prismshadow/penguin-server": "workspace:*",
"commander": "^13.0.0",
diff --git a/packages/core/package.json b/packages/core/package.json
index 841db4d7..5fa15fe3 100644
--- a/packages/core/package.json
+++ b/packages/core/package.json
@@ -65,7 +65,7 @@
"@penguinharness/use-bento-slides": "workspace:*",
"@penguinharness/use-firecrawl": "workspace:*",
"@penguinharness/use-claude-code": "workspace:*",
- "@prismshadow/agenthub": "^0.4.8",
+ "@prismshadow/agenthub": "^0.4.9",
"arktype": "^2.2.3",
"smol-toml": "^1.3.0",
"yaml": "^2.5.0"
diff --git a/pnpm-lock.yaml b/pnpm-lock.yaml
index 93b69078..951d5029 100644
--- a/pnpm-lock.yaml
+++ b/pnpm-lock.yaml
@@ -85,8 +85,8 @@ importers:
specifier: workspace:*
version: link:../../plugins/use-firecrawl
'@prismshadow/agenthub':
- specifier: ^0.4.8
+ specifier: ^0.4.9
'@prismshadow/penguin-core':
specifier: workspace:*
version: link:../core
@@ -164,8 +164,8 @@ importers:
specifier: workspace:*
version: link:../../plugins/use-firecrawl
'@prismshadow/agenthub':
- specifier: ^0.4.8
+ specifier: ^0.4.9
arktype:
specifier: ^2.2.3
version: 2.2.3
@@ -1129,8 +1129,8 @@ packages:
engines: {node: '>=18'}
hasBin: true
- '@prismshadow/[email protected]':
- resolution: {integrity: sha512-d+NBK/KaJeAA0kICkT1Ww4yVOf5TwuENsHXAxPp4LHMKL9CDIQotK9Wcb9jVa5WUjVEqLg/5Tiu/MhmlW7IW8A==}
+ '@prismshadow/[email protected]':
+ resolution: {integrity: sha512-Sjdm7wDDMhkKoW0v0ct3LgH6JedpnnGWwJ+NdEw8VjoXunwzLT39qAVKk/CVMizvoTUFEOgWshh6EGDnzODScA==}
'@prismshadow/penguin-cli@file:packages/cli':
resolution: {directory: packages/cli, type: directory}
@@ -4806,7 +4806,7 @@ snapshots:
dependencies:
playwright: 1.61.1
dependencies:
'@anthropic-ai/bedrock-sdk': 0.26.4([email protected])
'@anthropic-ai/sdk': 0.91.1([email protected])
@@ -4821,7 +4821,7 @@ snapshots:
- ws
- zod
dependencies:
'@anthropic-ai/bedrock-sdk': 0.26.4([email protected])
'@anthropic-ai/sdk': 0.91.1([email protected])
@@ -4851,7 +4851,7 @@ snapshots:
'@penguinharness/use-bento-slides': file:plugins/use-bento-slides
'@penguinharness/use-claude-code': file:plugins/use-claude-code
'@penguinharness/use-firecrawl': file:plugins/use-firecrawl
- '@prismshadow/agenthub': 0.4.8([email protected])([email protected])
+ '@prismshadow/agenthub': 0.4.9([email protected])([email protected])
'@prismshadow/penguin-core': file:packages/core([email protected])([email protected])
'@prismshadow/penguin-server': file:packages/server([email protected])
commander: 13.1.0
@@ -4882,7 +4882,7 @@ snapshots:
'@penguinharness/use-bento-slides': file:plugins/use-bento-slides
'@penguinharness/use-claude-code': file:plugins/use-claude-code
'@penguinharness/use-firecrawl': file:plugins/use-firecrawl
- '@prismshadow/agenthub': 0.4.8([email protected])([email protected])
+ '@prismshadow/agenthub': 0.4.9([email protected])([email protected])
arktype: 2.2.3
smol-toml: 1.6.1
yaml: 2.9.0
@@ -4909,7 +4909,7 @@ snapshots:
'@penguinharness/use-bento-slides': file:plugins/use-bento-slides
'@penguinharness/use-claude-code': file:plugins/use-claude-code
'@penguinharness/use-firecrawl': file:plugins/use-firecrawl
- '@prismshadow/agenthub': 0.4.8([email protected])([email protected])([email protected])
+ '@prismshadow/agenthub': 0.4.9([email protected])([email protected])([email protected])
arktype: 2.2.3
smol-toml: 1.6.1
yaml: 2.9.0
diff --git a/pnpm-workspace.yaml b/pnpm-workspace.yaml
index 3c60055c..b790567e 100644
--- a/pnpm-workspace.yaml
+++ b/pnpm-workspace.yaml
@@ -32,9 +32,9 @@ allowBuilds:
# targets NSIS only, so the script is deliberately not run.
electron-winstaller: false
# pnpm 11 withholds freshly published versions by default (supply-chain gate), which
-# would resolve agenthub back to 0.4.7 and lose the GLM image input that the glm-5.3-flash
-# preset's vision flag rests on. This package is first-party, so the exemption carries no
-# third-party risk.
-# Remove the entry at the next agenthub bump — by then 0.4.8 is long past the age gate.
+# would resolve agenthub back to the previous release and lose the fix the bump is for.
+# This package is first-party, so the exemption carries no third-party risk.
+# Every bump lands inside the age gate, so the entry names the pinned version and moves
+# with each bump rather than being removed.
minimumReleaseAgeExclude:
- - "@prismshadow/[email protected]"
+ - "@prismshadow/[email protected]"