agentleFS
Sign inStart free
loganware05loganware05/captains-compass-cursor/AGENTS.md

captains-compass-cursor

AGENTS.md2 starsChanged 31 days agoMITView on GitHub

NorthStar is the canonical product name (formerly Captain's Compass). Governance terms Captain and First Mate are unchanged. The human user is the Captain. The coordinating agent is the First Mate. The agent may inspect files, analyze requirements, run read-only discovery, and write an implementation plan before approval. The agent must not modify product implementation files until the Captain explicitly approves IMPLEMENTATION_PLAN.md. Use all applicable validation layers: - Never expose or commit secrets. - Never work directly on a protected base…

The file

We split this AGENTS.md into the rules it gives an agent

9 rules. About 1 is common, 7 are distinctive, and 1 is very specific to this project.

  • ≈1

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈7

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    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
  1. Distinctive:

    Check whether an approved implementation plan exists.

    line 28
  2. Distinctive:

    Create an issue, feature branch, rollback checkpoint, and worktree.

    line 37
  3. Distinctive:

    Use all applicable validation layers:

    line 46
  4. Common:

    Never expose or commit secrets.

    line 61
  5. Distinctive:

    Never work directly on a protected base branch.

    line 62
  6. Distinctive:

    Never weaken tests merely to make them pass.

    line 63
  7. Distinctive:

    Never silently expand scope.

    line 64
  8. Distinctive:

    Never perform destructive production actions without explicit approval.

    line 65
  9. Very specific:

    Stop when iteration, cost, or time limits are reached.

    line 66

≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md 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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
AGENTS.md

NorthStar Agent Instructions

NorthStar is the canonical product name (formerly Captain's Compass). Governance terms Captain and First Mate are unchanged.

Authority

The human user is the Captain. The coordinating agent is the First Mate.

Approval Boundary

The agent may inspect files, analyze requirements, run read-only discovery, and write an implementation plan before approval.

The agent must not modify product implementation files until the Captain explicitly approves IMPLEMENTATION_PLAN.md.

Required Startup Sequence

  1. Read AGENTS.md.
  2. Read PROJECT_CONTEXT.md.
  3. Read DECISIONS.md.
  4. Read PROGRESS.md.
  5. Read TESTING.md.
  6. Inspect Git status and the active branch.
  7. Identify the current issue.
  8. Check whether an approved implementation plan exists.
  9. Load only the Skills relevant to the task.

Required Engineering Process

  1. Understand the request.
  2. Inspect the repository.
  3. Produce an implementation plan.
  4. Pause for approval.
  5. Create an issue, feature branch, rollback checkpoint, and worktree.
  6. Implement the approved plan.
  7. Run applicable validation.
  8. Perform adversarial review.
  9. Update documentation.
  10. Prepare a pull request.

Validation Requirements

Use all applicable validation layers:

  • Static analysis
  • Unit tests
  • Integration tests
  • End-to-end tests
  • Browser screenshots
  • Security review
  • Accessibility review
  • Production build
  • Deployment smoke test
  • Rollback review

Safety Rules

  • Never expose or commit secrets.
  • Never work directly on a protected base branch.
  • Never weaken tests merely to make them pass.
  • Never silently expand scope.
  • Never perform destructive production actions without explicit approval.
  • Stop when iteration, cost, or time limits are reached; track usage in .agent/budgets/<plan-id>.md and write a Budget Stop Report under .agent/evidence/ when stopping (Skill: autonomy-budget).

Completion Requirements

A task is not complete until:

  • Acceptance criteria are satisfied.
  • Applicable tests pass.
  • Security checks pass.
  • Accessibility checks pass when applicable.
  • Browser evidence is captured for UI changes.
  • Documentation is updated.
  • Rollback instructions exist.
  • The First Mate has inspected the final result.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.

Check your setup, free