fix(server): isolate Cursor metadata generation

[saphid]

Cursor starts metadata requests in the project directory even though their prompts already contain the required input. This exposes title, branch-name, commit-message and PR-text generation to project files and instructions they do not need.

Each request now uses an empty temporary workspace, following the Claude change in #4169. Relative executable paths resolve from the original project. The workspace is removed when the ACP request scope closes, including cancellation; cleanup errors are logged without discarding generated output. This is workspace isolation, not an OS sandbox.

Verified after integrating upstream main 35be904, using Node 24.13.1:

• CursorTextGeneration and CursorAcpSupport: 13 tests passed, including isolation, relative executable resolution, cancellation and cleanup failure.
• Server typecheck and targeted lint/format passed.

Tests use the mock ACP agent; no live Cursor request or Windows runtime was exercised. No visual UI change.

Fresh independent read-only GPT-6.1 Sol review in the Codex harness found no actionable findings in the complete contribution. The reviewer independently repeated all 13 focused tests.

Earlier preparation: Claude Opus 5 in Claude Code via T3 Code. Current integration and verification: GPT-6 Astra in the Codex harness, via T3 Code.

Independent review: GPT-6.1 Sol in the Codex harness via T3 Code.

[cursor]

Cursor Bugbot has reviewed your changes using default effort and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 72afcbf1b8561fb4609b6de49c8f067bb03957d7. Configure here.}

[macroscopeapp]

Approvability

Verdict: Not approved

Macroscope's review found this PR not approvable — The production change alters every Cursor metadata request by changing its workspace and prompt handling, with tests covering isolation and cleanup. The PR also adds a line-level static-analysis suppression in the test suite, so human review is required.

^{You can add or adjust custom eligibility rules. Learn more.}

[macroscopeapp]

Macroscope has since reviewed this pull request. An earlier review was skipped by a cost limit; a review has now completed, so that notice no longer applies.

[cursor]

This PR is too large for Bugbot to review. It changes 99,173 lines and 4,020,811 characters. Split the change into smaller pull requests to get a review.

[coderabbitai]

Navigate logical layers of code changes, visualize relationships, and explore their blast radius.

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Repository: pingdotgg/t3code/.coderabbit.yaml

Review profile: CHILL

Plan: Advanced

Run ID: dda1f290-22ed-44b1-bc40-28fcb98e272f

📥 Commits

Reviewing files that changed from the base of the PR and between d6e595c7c1684daeadb8511a39a584f955c76b9a and c0d980a.

📒 Files selected for processing (1)

• apps/server/src/textGeneration/CursorTextGeneration.test.ts

Included review availability: Your plan provides up to 10 included reviews per hour; 6 remain after this review.

---

📝 Walkthrough

Walkthrough

Cursor ACP generation now runs in a temporary metadata workspace. Relative binary paths resolve from the source directory. Prompts restrict processing to the supplied input and request JSON-only output. Tests cover workspace isolation, cancellation cleanup, and cleanup failures.

Changes

Cursor metadata isolation

Layer / File(s)	Summary
Metadata workspace lifecycle apps/server/src/textGeneration/CursorTextGeneration.ts	runCursorJson creates and removes a temporary metadata workspace. Relative Cursor binary paths resolve against the source directory. Generation call sites pass sourceCwd.
Prompt and workspace assertions apps/server/src/textGeneration/CursorTextGeneration.ts, apps/server/src/textGeneration/CursorTextGeneration.test.ts	The prompt restricts tools, file access, commands, and questions, and requests only the specified JSON object. Tests verify workspace isolation, relative binary handling, ACP options, and prompt wording.
Cancellation and cleanup validation apps/server/scripts/acp-mock-agent.ts, apps/server/src/textGeneration/CursorTextGeneration.test.ts	The mock agent records prompt start at a configured marker path. Tests check workspace removal after cancellation and generated metadata preservation when cleanup fails.

Priority: ➖ Normal

Estimated code review effort: 3 (Moderate) | ~20 minutes

Change: Bug fix

Sequence Diagram(s)

sequenceDiagram
  participant TextGeneration
  participant FileSystem
  participant CursorACP
  TextGeneration->>FileSystem: Create temporary metadata workspace
  TextGeneration->>CursorACP: Start with isolated cwd and JSON prompt
  CursorACP-->>TextGeneration: Return generated metadata
  TextGeneration->>FileSystem: Remove workspace on release

Loading

Merge Risk: ⚪ Minimal · up to c0d98

The workspace isolation change has no identified merge-blocking issue after normal checks.

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 1 functions across 3 files.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Title check	✅ Passed	The title clearly and concisely describes the main change: isolating Cursor metadata generation in the server.
Description check	✅ Passed	The description explains what changed, why it changed, cleanup behavior, scope limitations, testing, and the absence of UI changes. It does not use the template headings or include the checklist, but …

• Fix all pre-merge checks with AI

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create a new PR

---

_{Comment @coderabbitai help to get the list of available commands.}

[saphid]

Independent automated first-pass review (Amp/Astra agent mode) of head de8aa56ba against upstream/main: no findings. Correctness, invariants, edge cases, Effect conventions, and performance all checked; temp-workspace scoping, relative binary-path resolution, and cleanup-on-failure semantics verified sound against the source. Nothing to address, no commits pushed.

[saphid]

Friendly review nudge @juliusmarminge @maria-rcks — this is mergeable and hasn't had a maintainer pass yet. Independent bot/agent reviews have run with findings triaged in-commit (see receipts in earlier comments). Full queue context and status: #10688.

[juliusmarminge]

Thanks for working on this. We merged the orchestrator V2 rewrite in #2829, and we are closing this PR as part of that transition.

The patch conflicts with the rewrite in apps/server/scripts/acp-mock-agent.ts, apps/server/src/textGeneration/CursorTextGeneration.ts. Even where the conflict is small enough to rebase, we are asking for fresh PRs against the new base so we can review and verify the behavior in V2.

Sorry for the extra work this creates. If the change is still needed on V2, please rebuild it on current main, verify it there, and open a new PR linking back here. We're closing the current implementation without assuming the underlying request is resolved.