fix(cursor): preserve Grok wire model prefix

[jonathanli12]

Summary

• preserve Cursor's required cursor- prefix for regular Grok 4.5 outbound wire IDs
• keep discovery matching on canonical prefix-free IDs
• add low, medium, and high coverage plus a discovery-to-request invariant

Validation

• focused Cursor tests: 46 passed
• typecheck: passed
• privacy scan: passed
• full suite: 9,457 passed, 8 skipped, one unrelated timeout; isolated rerun passed 20 tests
• git diff check: passed

Summary by CodeRabbit

• Bug Fixes

  • Corrected Cursor Grok 4.5 request identifiers for low, medium, and high effort levels.
  • Preserved the expected request format for Grok 4.5 Fast, including its separate effort and fast parameters.
  • Improved compatibility with live model discovery identifiers across supported effort levels.

• Documentation

  • Clarified the different model ID and effort parameter formats used by regular and fast Grok 4.5 tiers.

Review readiness checklist

This PR stays in draft until every box below is ticked. Tick all four boxes once the requirements are met:

• All CI tests are green on my local testing.

• I pushed my PR to the latest dev commit.

• I resolved all correct Codex and CodeRabbit findings.

• My PR is ready for review.

[coderabbitai]

📝 Walkthrough

Walkthrough

Cursor now sends cursor--prefixed effort IDs for regular Grok 4.5 requests. The request builder uses a dedicated mapping helper. Tests verify low, medium, and high IDs against recorded discovery data. Documentation distinguishes regular and fast Grok 4.5 behavior.

Changes

Cursor Grok 4.5 effort ID handling

Layer / File(s)	Summary
Effort-qualified request ID mapping src/adapters/cursor/effort-map.ts, src/adapters/cursor/request-builder.ts	cursorRequestWireModelIdWithEffort prefixes regular grok-4.5 effort IDs with cursor-. The request builder imports and calls the new helper.
Discovery ID validation and documentation tests/cursor-effort-suffix.test.ts, docs-site/src/content/docs/reference/adapters.md	Tests cover low, medium, and high discovery IDs. Documentation describes prefixed regular IDs and separate fast-model parameters.

Estimated code review effort: 2 (Simple) | ~10 minutes

Possibly related PRs

• lidge-jun/opencodex#646: Modifies Cursor effort-tier mapping and related tests.
• lidge-jun/opencodex#730: Modifies Cursor effort-based wire model ID construction.
• lidge-jun/opencodex#881: Modifies Cursor Grok 4.5 effort mapping, request construction, documentation, and tests.

Suggested labels: bug

Suggested reviewers: lidge-jun, wibias, ingwannu

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
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.
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly and concisely describes the main change: preserving the Cursor wire model prefix for Grok requests.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

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

[coderabbitai]

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@tests/cursor-effort-suffix.test.ts`:
- Around line 122-136: The test named “regular grok-4.5 requests exactly match
live prefixed discovery ids” uses hard-coded values rather than discovery data
and cannot validate the exact wire ID. Rename it as a static mapping test, or
replace liveIds with a clearly sourced discovery/recorded fixture and assert the
discovery-to-request invariant separately from isCursorModelAvailableForAccount.

🪄 Autofix

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: adea0bce-25ed-44f9-8a8b-44d0ca59bcf7

📥 Commits

Reviewing files that changed from the base of the PR and between b39eecf and bb0aa9d.

📒 Files selected for processing (4)

• docs-site/src/content/docs/reference/adapters.md
• src/adapters/cursor/effort-map.ts
• src/adapters/cursor/request-builder.ts
• tests/cursor-effort-suffix.test.ts

[github-actions]

⏳ DRAFT

• review readiness checklist open (0/4 boxes ticked).

What to do

• Tick all four boxes in the PR description once you're done (currently 0/4).

Review readiness checklist

• ⬜ All CI tests are green on my local testing.
• ⬜ I pushed my PR to the latest dev commit.
• ⬜ I resolved all correct Codex and CodeRabbit findings.
• ⬜ My PR is ready for review.

0/4 boxes ticked.

This PR stays in draft until every box above is ticked.

[github-actions]

✅ Deterministic PR hygiene checks passed.

[lidge-jun]

Closing in favor of #1208, which adopts this change as-is.

To be clear about why: this PR needed no changes. It was independently re-audited against current dev and confirmed correct — the request-only helper is the right seam, discovery normalization is properly left alone, and Grok Fast keeps its parameterized form. We are rebuilding it on our stack per the sequential-PR convention, not because anything was wrong with it. You are credited in the commit.

Worth saying plainly: this sat unmergeable because 524 workflow runs were queued awaiting maintainer approval, 39 of them on open-PR branches including this one. The readiness gate verifies the ci check, and a run that was never allowed to start has no result to verify — so this could not leave draft regardless of what you did. That was our backlog, not your PR.

Two notes carried into #1208: the fixture pins against recorded discovery output rather than live Cursor state, so it catches mapper drift but not a future rename without a manual refresh; and the change prefixes only regular Grok 4.5, so it does not address #1162.

Thanks — this was a clean fix.