Cut off Anthropic Messages stream after 5 min max duration (#321432)

[meganrogge]

No description provided.

[meganrogge]

/requires-eval-assessment terminalbench2 gpt-5.5,claude-opus-4.6,claude-opus-4.7 vscode-copilot-cli

[vs-code-engineering]

⏳ Queued vscode build for d948744512738e616795dc36f138b3bb9fafae40 (step 1/2).

• Build: https://dev.azure.com/monacotools/Monaco/_build/results?buildId=448323
• When this succeeds, the eval-assessment publish build will be queued automatically.

[Copilot]

Pull request overview

This PR adds a hard maximum duration for Anthropic Messages API streaming responses in the Copilot extension’s node endpoint implementation, preventing requests from hanging indefinitely when the server returns 200 and then stops producing SSE chunks.

Changes:

• Introduces a module-level constant defining a 5-minute maximum streaming duration.
• Adds a timer-based cutoff that rejects the returned AsyncIterable and destroys the streaming body when the limit is exceeded.
• Ensures the cutoff timer is cleared when streaming finishes (successfully or not).

Show a summary per file

File	Description
extensions/copilot/src/platform/endpoint/node/messagesApi.ts	Adds a max-duration cutoff for Anthropic Messages SSE streaming to avoid indefinite hangs.

Copilot's findings

• Files reviewed: 1/1 changed files
• Comments generated: 2

[vs-code-engineering]

⏳ Queued vscode build for 878695a89d12a1e78839921fec2b12e4a9bcbb99 (step 1/2).

• Build: https://dev.azure.com/monacotools/Monaco/_build/results?buildId=448336
• When this succeeds, the eval-assessment publish build will be queued automatically.

[vs-code-engineering]

🚀 Queued eval-assessment publish build for d36275542e98716b882a87908aec1b37e875da58 (step 2/2).

• Pipeline run: https://dev.azure.com/monacotools/Monaco/_build/results?buildId=448361
• On success, publishes @vscode/vscode-copilot-evaluation-agent@0.0.0-dev.d36275542e on the dev tag.

[vs-code-engineering]

🔬 Queued eval-assessment benchmark for 7cc2487963.

• Package: @vscode/vscode-copilot-evaluation-agent@0.0.0-dev.d36275542e (dev tag)
• Agent: vscode-copilot-cli
• Benchmark: terminalbench2
• Tracking issues:
  • terminalbench2 / gpt-5.5: https://github.com/github/evald/issues/33820
  • terminalbench2 / claude-opus-4.6: https://github.com/github/evald/issues/33821
  • terminalbench2 / claude-opus-4.7: https://github.com/github/evald/issues/33822

Results will be posted back here when the run completes.

[vs-code-engineering]

✅ Eval-assessment build published.

• Package: @vscode/vscode-copilot-evaluation-agent@0.0.0-dev.d36275542e (tag: dev)
• Install: npm install @vscode/vscode-copilot-evaluation-agent@0.0.0-dev.d36275542e
• Pipeline run: https://dev.azure.com/monacotools/Monaco/_build/results?buildId=448361