feat(integrations): Cursor + install.sh --dry-run + stale-doc sweep - #32
Merged
Conversation
Phase 1 of the post-1.0 roadmap. No wire-format change; all additive.
integrations/cursor/ — mirrors integrations/claude-code/ shape:
- install.sh registers ~/.cursor/mcp.json globally and optionally
copies .cursor/rules/{crumb-export,crumb-import,crumb-it}.mdc
into the current project. Marker-tracked, idempotent.
- uninstall.sh respects the install.sh marker (preserves
user-managed entries).
- mcp.json.template uses the same __CRUMB_PYTHON__ /
__CRUMB_INSTALL_PATH__ substitution shape as Claude Code.
- Three .mdc rule files: crumb-it is alwaysApply: true (ambient
format recognition); crumb-export and crumb-import fire on
demand or on .crumb file open.
install.sh --dry-run on both Claude Code and Cursor installers.
Shows every action (mkdir, file write, MCP merge, prompt) without
touching the filesystem. Reject unknown args with exit 2.
Stale design-doc sweep:
- docs/v1.2-ref-resolution.md, docs/v1.2-fold-heuristic.md:
cleared TODO(author) markers; "shipped in v=1.3" headers.
- docs/v1.3-DRAFT.md: "shipped" header.
- docs/v1.4-scoping.md: "shipped" header.
- docs/v1.4/{agent-failure-modes,handoff-deadlines,typed-checks}.md:
"shipped in v=1.4" headers.
- docs/integrations/cursor.md: "Plan only" → "shipping".
- grep -rln "TODO(author)" docs/ now returns zero.
README: new "Native integrations" section with one-line install
pointers for Claude Code and Cursor; CLI install block now mentions
crumb hello and crumb doctor.
21 new tests in tests/test_integrations_smoke.py:
- per-integration installer syntax + --help + --dry-run + unknown-arg
- MCP template renders to valid JSON after substitution
- Cursor rule frontmatter checks (crumb-it always-apply, others not)
- Claude Code slash command frontmatter checks
- uninstaller marker-check assertion
673 passing total (was 652).
https://claude.ai/code/session_01Ewf1noyQc9gc6DmohyR32z
Contributor
✅ Shadow AI ScanNo Shadow AI findings detected at or above medium risk. |
Contributor
CRUMB Bench Results
Powered by crumb-format — two-stage context compression |
1 similar comment
Contributor
CRUMB Bench Results
Powered by crumb-format — two-stage context compression |
slavazeph-coder
marked this pull request as ready for review
May 8, 2026 16:28
5 tasks
slavazeph-coder
added a commit
that referenced
this pull request
May 8, 2026
First post-1.0 minor. No wire-format change (stays at v=1.4). Semver-minor: adds Cursor integration (PR #32), three top-level CLI commands (PR #31: hello, doctor, friendlier argparse errors), --dry-run on installers, stale-doc cleanup. CLI_VERSION → 1.1.0. pyproject.toml version → 1.1.0. 673 tests passing. 39/39 examples + fixtures green on Python and Node.
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: dfc3880925
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
Comment on lines
+213
to
+214
| echo "$CRUMB_MANAGED_MARKER" | ||
| cat "$src" |
There was a problem hiding this comment.
Do not prepend the managed-marker line before copying .mdc rules, because this makes installed rule files start with <!-- ... --> instead of ---. The integration’s own rule contract assumes frontmatter is at the top (see tests/test_integrations_smoke.py asserting text.startswith("---\n")), so a non-dry-run install can write rule files Cursor may not parse/apply, breaking crumb-export/crumb-import/crumb-it behavior for end users.
Useful? React with 👍 / 👎.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Phase 1 of the post-1.0 roadmap. Doubles the IDE coverage (Claude Code + Cursor today) and lays trust groundwork (
--dry-runpreviews + integration smoke tests + stale-doc cleanup). No wire-format change; all additive.What ships
integrations/cursor/— Cursor is now a one-line installbash <(curl -fsSL https://raw.githubusercontent.com/XioAISolutions/crumb-format/main/integrations/cursor/install.sh)install.sh~/.cursor/mcp.json; optionally copies rule files into./.cursor/rules/. Idempotent, marker-tracked.uninstall.shmcp.json.template__CRUMB_PYTHON__/__CRUMB_INSTALL_PATH__substitution shape as Claude Code. Carries_managed_byfield.rules/crumb-export.mdckind=task(or right kind) crumb.rules/crumb-import.mdc.crumbfile is opened or pasted.rules/crumb-it.mdcalwaysApply: true— ambient format recognition for the verbal trigger.README.mdMirrors the Claude Code integration's three-artifact pattern. Cursor uses
.cursor/rules/*.mdcfiles instead of slash commands; the per-project rules install only when the user is in a.cursor/-tracked dir or opts in interactively.install.sh --dry-runBoth
claude-code/install.shandcursor/install.shaccept--dry-run(or-n). Every write —mkdir, file write, MCP merge, CLAUDE.md append prompt — prints what it would do without touching the filesystem.--helpand unknown-arg rejection (exit 2) also added.Stale-doc sweep
Eight docs that described draft-or-shipped designs are now correctly labeled.
grep -rln "TODO(author)" docs/returns zero.docs/v1.2-ref-resolution.md## TODO(author)docs/v1.2-fold-heuristic.md## TODO(author)docs/v1.3-DRAFT.mddocs/v1.4-scoping.mddocs/v1.4/handoff-deadlines.mddocs/v1.4/typed-checks.mddocs/v1.4/agent-failure-modes.mddocs/integrations/cursor.mdREADME
Added a "Native integrations —
crumb itinside your AI tool" section right after the CLI install block, with one-line install commands for both Claude Code and Cursor and pointers to the per-integration READMEs. Existing CLI block now mentionscrumb helloandcrumb doctorso new users see them.Test plan
python3 -m pytest tests/ -q— 673 passing (was 652; +21 intests/test_integrations_smoke.py)bash -n integrations/{claude-code,cursor}/install.sh integrations/{claude-code,cursor}/uninstall.sh— all syntax-cleanbash integrations/cursor/install.sh --dry-runproduces clean preview, writes nothingbash integrations/claude-code/install.sh --dry-rundittobash integrations/{claude-code,cursor}/install.sh --bogusexits 2 with clear error.mdcrules pass frontmatter validation;crumb-itisalwaysApply: true, othersfalse_managed_bymarker before deletinggrep -rln "TODO(author)" docs/returns nothingStats
19 files changed, +807 / -41.
integrations/cursor/(new)integrations/claude-code/install.sh--dry-run,--help, unknown-arg rejectiontests/test_integrations_smoke.py(new)docs/README.mdCHANGELOG.mdOut of scope (for follow-ups)
docs/integrations/.tests/test_e2e_roundtrip.py) — Phase 2 work. This PR's smoke tests cover installer shape; e2e coversnew → handoff → receive → resume.crumb contextglow-up — Phase 2 work.Sequencing
Stacks cleanly on top of
mainpost-PR-31. Phase 2 (e2e tests + JSON schema +crumb context) can start once this lands.https://claude.ai/code/session_01Ewf1noyQc9gc6DmohyR32z
Generated by Claude Code