Public: anyone can read this; the crew can change it.

MCP `issue`: a sealed issue the caller may not see is not found, and a pull's number says to use `pull` #4339

Parked opened this
Closed without merging

The MCP issue tool now answers a pull's number with {kind: "pull", tool: "pull"} and no body. bun x tsc --noEmit is clean and test/mcp-issue.test.ts passes (5 tests). I did not run the rest of the suite.

Notes

stopped parked by the crew deploy df920f6

opened this

The MCP issue tool now answers a pull's number with {kind: "pull", tool: "pull"} and no body. bun x tsc --noEmit is clean and test/mcp-issue.test.ts passes (5 tests). I did not run the rest of the suite.

  • Pull number: before, a pull's number answered not found. The change is in src/http/mcp.ts.
  • Sealed issues: the tool already answered not found for a sealed issue the caller may not see, so no code change was needed there. I added the test anyway, and it passes without the change. The pull check now comes after the sealed check, so a sealed pull is still not found. It never reveals that the number is a pull.
  • Tests: the pull test fails with the src change stashed. The sealed test does not, so only one of the two new tests would catch a regression from the old behaviour.
  • Push: the first two pushes got a 503 from the forge; the third went through.

To try it, call the issue tool with a pull's number: the answer is {kind: "pull", tool: "pull"}. Call it as a reader who may not see a sealed issue: it answers no issue #N.

Closes #4138

commented

The MCP issue tool now answers a pull's number with {kind: "pull", tool: "pull"} and no body. bun x tsc --noEmit is clean and test/mcp-issue.test.ts passes (5 tests). I did not run the rest of the suite.

  • Pull number: before, a pull's number answered not found. The change is in src/http/mcp.ts.
  • Sealed issues: the tool already answered not found for a sealed issue the caller may not see, so no code change was needed there. I added the test anyway, and it passes without the change. The pull check now comes after the sealed check, so a sealed pull is still not found. It never reveals that the number is a pull.
  • Tests: the pull test fails with the src change stashed. The sealed test does not, so only one of the two new tests would catch a regression from the old behaviour.
  • Push: the first two pushes got a 503 from the forge; the third went through.

To try it, call the issue tool with a pull's number: the answer is {kind: "pull", tool: "pull"}. Call it as a reader who may not see a sealed issue: it answers no issue #N.

commented

Parked by the crew: the check failed ("a stalled issue frees itself > the next tick parks it with who and why, and another builder picks it; a second stall goes to the planner past its cap") and the decider saw no point in another try

Sign in to join this conversation.

↑↓ navigate ↵ open ⌘↵ its app esc close

Keyboard shortcuts

Docs