agentHost: coordinate overlapping sessions

[Megan Rogge (meganrogge)]

Fixes #330292

Summary

• align Agent Host overlap coordination with the GitHub Copilot app workflow
• align Agent Host session coordination with a staged overlap workflow: identify the current session, discover active peers across the same project or repository, and exclude the current session
• compare session titles, activity, branches, pull requests, and changed files before loading conversation context
• inspect and message only plausible overlaps, agree on ownership or sequencing before editing shared files, and continue independently without waiting when work does not conflict
• clarify the overlap-relevant metadata returned by list_sessions and keep a short check-first reminder in create_session

[Copilot]

Pull request overview

Adds cross-session coordination guidance to prevent overlapping Agent Host work.

Changes:

• Adds universal session-coordination prompt guidance.
• Expands session tool descriptions.
• Updates prompt composition tests.

Show a summary per file

File	Description
toolInstructions.test.ts	Tests coordination prompt composition.
sessionServerTools.ts	Adds overlap checks to session tools.
toolInstructions.ts	Registers universal coordination guidance.

Review details

Tip

Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

• Files reviewed: 3/3 changed files
• Comments generated: 1
• Review effort level: Balanced

[Megan Rogge (meganrogge)]

/requires-eval-assessment vscbench gpt-5.5,claude-opus-4.7,claude-opus-4.8

sessionTarget: agent-host-copilotcli

[vs-code-engineering]

⏳ Queued vscode build for 635ebb8f6a11f82da5a95b43a0ed68898778baa3 (step 1/2).

• Build: https://dev.azure.com/monacotools/Monaco/_build/results?buildId=463721
• When this succeeds, the eval-assessment publish build will be queued automatically.

[vs-code-engineering]

🚀 Queued eval-assessment publish build for ff9e69f270673cb44bd28ccdab494e0f23e62088 (step 2/2).

• Pipeline run: https://dev.azure.com/monacotools/Monaco/_build/results?buildId=463741
• On success, publishes @vscode/vscode-copilot-evaluation-agent@0.0.0-dev.ff9e69f270 on the dev tag.

[vs-code-engineering]

❌ Eval-assessment build did not succeed (result: failed).

• Pipeline run: https://dev.azure.com/monacotools/Monaco/_build/results?buildId=463741