Skip to content

Evenfire is an intelligence layer platform for running LLM agents that take real actions, on your infrastructure

TypeScript

test(mcp-host): fail the native journey harness on swallowed assertions

Alfredo Lopez208db655d328
diff --git a/mcp-host/package-lock.json b/mcp-host/package-lock.json
index 0d8680d47..ae1cfe4e2 100644
--- a/mcp-host/package-lock.json
+++ b/mcp-host/package-lock.json
@@ -1,12 +1,12 @@
{
"name": "clerum-mcp-host",
- "version": "0.1.500",
+ "version": "0.1.501",
"lockfileVersion": 3,
"requires": true,
"packages": {
"": {
"name": "clerum-mcp-host",
- "version": "0.1.500",
+ "version": "0.1.501",
"license": "SEE LICENSE IN LICENSE",
"dependencies": {
"@anthropic-ai/sdk": "^0.32.0",
diff --git a/mcp-host/package.json b/mcp-host/package.json
index 07d66f086..5f434642c 100644
--- a/mcp-host/package.json
+++ b/mcp-host/package.json
@@ -1,6 +1,6 @@
{
"name": "clerum-mcp-host",
- "version": "0.1.500",
+ "version": "0.1.501",
"description": "MCP Host - reads Host CRD and provides LLM access via OpenAI/Claude",
"main": "dist/main.js",
"engines": {
diff --git a/mcp-host/src/core/orchestration/__tests__/nativeToolPresentation.integration.test.ts b/mcp-host/src/core/orchestration/__tests__/nativeToolPresentation.integration.test.ts
index 88be1edfd..8f3001e2c 100644
--- a/mcp-host/src/core/orchestration/__tests__/nativeToolPresentation.integration.test.ts
+++ b/mcp-host/src/core/orchestration/__tests__/nativeToolPresentation.integration.test.ts
@@ -197,6 +197,9 @@ async function runScenario(scenario: Scenario, turns: Turn[]) {
: scenario.mcp
)
const providerCalls: ProviderCall[] = []
+ // A failed assertion inside a turn callback throws inside the provider, which
+ // TaskExecutor turns into a task failure. Record it so the test fails with it.
+ const turnErrors: unknown[] = []
const usage = { input_tokens: 1, output_tokens: 1, total_tokens: 2 }
const provider: SingleTurnProvider = {
getProviderType: () => scenario.provider,
@@ -209,9 +212,15 @@ async function runScenario(scenario: Scenario, turns: Turn[]) {
completeSingleTurnWithTools: async (messages, tools) => {
const call = { messages: structuredClone(messages), tools: structuredClone(tools) }
providerCalls.push(call)
- const turn = turns[providerCalls.length - 1]
- if (!turn) throw new Error(`Unexpected provider call ${providerCalls.length}`)
- const reply = turn(call)
+ let reply: ReturnType<Turn>
+ try {
+ const turn = turns[providerCalls.length - 1]
+ if (!turn) throw new Error(`Unexpected provider call ${providerCalls.length}`)
+ reply = turn(call)
+ } catch (error) {
+ turnErrors.push(error)
+ throw error
+ }
return reply.tool_calls
? {
content: null,
@@ -282,8 +291,21 @@ async function runScenario(scenario: Scenario, turns: Turn[]) {
dynamicEnvProvider: () => ({}),
}
const executor = new TaskExecutor(task, deps)
+ /**
+ * Fails the test with the first turn-callback error, or if the task failed.
+ * Once the task has settled without suspending, every scripted turn must have
+ * been consumed. Called after `run()`, and again by tests after each resume.
+ */
+ const settle = () => {
+ if (turnErrors.length > 0) throw turnErrors[0]
+ expect(deps.onFail).not.toHaveBeenCalled()
+ if (executor.executorState !== 'waiting_approval') {
+ expect(providerCalls).toHaveLength(turns.length)
+ }
+ }
await executor.run()
- return { executor, deps, providerCalls, responses, workspaceService }
+ settle()
+ return { executor, deps, providerCalls, responses, workspaceService, settle }
}
/** Search → describe → bridged call for the pptx generator, then a final answer. */
@@ -388,6 +410,7 @@ describe('T1 — native discovery journeys through TaskExecutor.run()', () => {
expect(run.deps.onApprovalNeeded).toHaveBeenCalledTimes(1)
await run.executor.resumeAfterApproval(false)
+ run.settle()
expect(run.executor.executorState).toBe('completed')
expect(run.providerCalls).toHaveLength(2)
expect(
@@ -750,6 +773,40 @@ describe('T3 — negative guards with liveness witnesses (through TaskExecutor.r
expect(pptxFiles()).toEqual(['bridged-deck.pptx'])
})
+ it('a bridged native call passes the same parameter validation as a direct one', async () => {
+ const privateUrl = { url: 'http://127.0.0.1:9/internal', method: 'GET' }
+ const blocked = 'Parameter validation failed: Non-public target "127.0.0.1" is blocked'
+ const run = await runScenario(
+ {
+ provider: 'codex-subscription',
+ codexMode: 'direct',
+ native: 'auto',
+ mcp: 2,
+ approval: true,
+ },
+ [
+ () =>
+ toolCalls(
+ bridged('bridged-http', 'http_request', privateUrl),
+ { id: 'direct-http', name: 'http_request', arguments: privateUrl },
+ bridged('pptx-call', 'clerum__generate_pptx', PPTX_ARGS)
+ ),
+ ({ messages }) => {
+ expect(toolResult(messages, 'bridged-http')).toContain(blocked)
+ expect(toolResult(messages, 'direct-http')).toContain(blocked)
+ // Witness: a valid bridged native in the same batch executes.
+ expect(toolResult(messages, 'pptx-call')).toContain('File generated: bridged-deck.pptx')
+ return { content: 'done' }
+ },
+ ]
+ )
+ // http_request requires approval and the gate is live here (it suspends the
+ // same call when validation is skipped); a rejected call never reaches it.
+ expect(run.deps.onApprovalNeeded).not.toHaveBeenCalled()
+ expect(run.executor.executorState).toBe('completed')
+ expect(pptxFiles()).toEqual(['bridged-deck.pptx'])
+ })
+
it('the catalog never lists bridge tools while it lists the hidden generators', async () => {
let results: Array<{ name: string }> = []
await runScenario(