Master-Planner / rules
Cursor rule0 starsChanged 5 months agoMITView on GitHub
Prevents premature implementation and destructive actions during planning
- Deletes or force-pushes
- Installs packages
- Commits and pushes
Where files are headingNew Cursor rule files have more rules about planning and tracking tasks, and fewer about code style.
Of the Cursor rule files we read, the 4,183 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.
In this repository12 other files it gives its agents.
- .cursor/rules/000-harness-bootstrap.mdcCursor rule
- .cursor/rules/002-source-discipline.mdcCursor rule
- .cursor/rules/003-anti-rationalization.mdcCursor rule
- .cursor/rules/004-memory-and-resume.mdcCursor rule
- doubt-researchSkill
- intake-interviewSkill
- master-plannerSkill
- plan-grillerSkill
- project-memorySkill
- security-rollbackSkill
- spec-drafterSkill
- task-decomposerSkill
The file
We split this 001-planning-safety.mdc into the rules it gives an agent
5 rules. About 0 are common, 2 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- could be automatic: a setting could enforce it
- Very specific: line 11
No implementation before plan approval: Do not write, generate, or modify production source code (.js, .ts, .py, .go, .rs, .java, .rb, …
- Very specific: line 12
No destructive commands during planning: Do not execute rm -rf, DROP TABLE, git push --force, docker system prune, or any irreversible …
- Very specific: line 13
No unreviewed third-party installs: Do not run npm install, pip install, cargo add, npx, or any package manager command without explicit …
- Distinctive: line 14
No secrets in planning docs: Never write API keys, passwords, tokens, or PII into any planning artifact.
- Distinctive: line 23
When the agent drifts from the planning workflow, escalate gradually:
≈ “Common” is an estimate: at least 1 in 5 of the 6,668 Cursor rule files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
--- description: Prevents premature implementation and destructive actions during planning globs: * alwaysApply: true --- # Planning Safety ## Hard Blocks These are non-negotiable. The agent must refuse and explain why. 1. **No implementation before plan approval.** Do not write, generate, or modify production source code (`.js`, `.ts`, `.py`, `.go`, `.rs`, `.java`, `.rb`, `.swift`, `.kt`, etc.) unless `.cursor/plans/<project>/manifest.md` explicitly shows `Status: READY_FOR_DEV` or the user has explicitly approved the plan and requested implementation. 2. **No destructive commands during planning.** Do not execute `rm -rf`, `DROP TABLE`, `git push --force`, `docker system prune`, or any irreversible command during any planning phase. 3. **No unreviewed third-party installs.** Do not run `npm install`, `pip install`, `cargo add`, `npx`, or any package manager command without explicit user approval. Propose the install, explain why, and wait. 4. **No secrets in planning docs.** Never write API keys, passwords, tokens, or PII into any planning artifact. ## Permitted During Planning - Example code in markdown fenced blocks (for illustration, not execution). - Research commands (`curl` to test a public API, `npx --help` to check a tool's interface). - File reads to understand existing codebase. - Creating/updating `.cursor/plans/` artifacts. ## Soft Enforcement Escalation When the agent drifts from the planning workflow, escalate gradually: - **Silent**: Agent is following the plan. No intervention needed. - **Hint**: "Note: we haven't completed the spec yet. Want to finish that first?" - **Warning**: "Warning: implementation is being attempted before the plan is approved." - **Hard stop**: Refuse the action and cite this rule. ## Anti-Rationalization | Excuse | Reality | |--------|---------| | "The change is trivial, no plan needed" | If it is tiny and implementation was explicitly requested, proceed without harness planning. If design decisions exist, create a lightweight plan. | | "The user seems impatient" | A bad plan wastes more time than a quick spec. | | "I can plan and code simultaneously" | This is how drift starts. Finish the plan first. | | "The plan is obvious, I'll just start" | For non-tiny work, write the obvious plan. It takes 30 seconds. |
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.