agentleFS
mehrad-dmmehrad-dm/mastermind/skills/constitution/SKILL.md

constitution

Skill23 starsChanged 2 months agoMITView on GitHub

Use when a project needs rules every feature must obey, or those rules change: \"set up our principles\", \"write a constitution\", \"these are our non-negotiables\", \"from now on every feature must...\", \"we never allow X in this codebase\", \"amend the rules\", starting spec-first work on a project with no specs/constitution.md. Not for MasterMind's own engineering defaults, and not for one feature's requirements (that's `specify`).

6 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

1 rule. 0 are common, 0 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 4 edited versions we read keep these.

  • 0

    Distinctive

    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
  1. Very specific:

    Start from the template: Use specs/.templates/constitution.md if it exists, else template.md beside this file.

    line 16

“Common” is measured: kept by at least 6 in 10 of the 4 edited versions of this file we could read. “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
SKILL.md

MasterMind: Constitution

A constitution is the short list of rules this project refuses to break, written so a reviewer can check a change against each one. It outranks every spec, plan and task (~/.mastermind/engineering/core/spec-driven.md). Request: $ARGUMENTS.

Write it

  1. Read what the project already says. README, CI config, lint and type settings, .mastermind/brief.md, existing specs/. Rules the repo already enforces are facts to record, not questions to ask.
  2. Start from the template. Use specs/.templates/constitution.md if it exists, else template.md beside this file. Write to specs/constitution.md.
  3. Draft at most seven principles. One is enough when the project has one rule. Each is a name, a few MUST or MUST NOT statements, and one line of rationale. More than seven means some are preferences.
  4. Ask only for what the repo cannot tell you. One question at a time, each with your recommended answer (interview). Privacy stance, compliance, and what the product will never do are the usual ones.
  5. Leave no placeholder behind. A value nobody knows yet stays as TODO(<what>): <why it is unknown> and is listed in your report.

Amend it

  • Version every change in the footer line, semver:
    • MAJOR: a principle is removed or redefined so that old work may now violate it.
    • MINOR: a principle or section is added, or its guidance materially changes without making old work non-compliant, which includes narrowing an obligation.
    • PATCH: wording, typos, clarifications that change no obligation.
  • Record the amendment as a comment block at the top of the file: the version change, what changed and why, the date, and who decided it.
  • Ripple the change. List every living doc, open spec, plan and template that now conflicts with the new rule. Update each one, or name it in the comment block as pending, with the reason.

Write rules a reviewer can check

Weak:   Code should be well tested.
Strong: Money arithmetic MUST have a failing test written before the code that makes it pass.
        Rationale: rounding bugs in payouts are invisible until a customer finds them.

Ratchet: Test coverage of src/billing MUST NOT fall below 81.4% (measured 2026-10-07 by `npm run cov`).
        Rationale: we cannot reach 100% today, and we refuse to get worse.

A number you cannot measure today is invented. Record where the project is now, with the command that measured it, and forbid going backwards. Raise the recorded value when the project improves. For a product built on a model, one rule is usually worth writing down: a guardrail in a prompt is not a security boundary, so what the model's output may trigger is limited in code.

A rule nobody can check produces an argument in every review. SHOULD is allowed only with the condition under which it may be skipped, stated in the same line.

Gotchas

  • Do not copy MasterMind's defaults in. "Verify before done" and "no lazy placeholders" already hold everywhere. The constitution carries what is true of this project only.
  • Do not let a feature request sneak in. "We need dark mode" is a spec. "Our surfaces are dark only" is a rule. Requests that arrive here go to specify, named in your report as deferred.
  • Existing code that already breaks a new rule is known debt, not a crisis. List it in the amendment block with its location (~/.mastermind/engineering/core/spec-driven.md, Code that predates the rules).
  • The constitution never edits code. Amending a rule that existing code now breaks produces findings for converge, not a silent refactor.
  • A team's rules are the team's decision. An unconfirmed draft is version 0.1.0. It becomes 1.0.0 on one real confirmation. A rule the user dictated in their own words is already confirmed; only what you inferred or proposed needs it. A rule the user never agreed to is a rule they will route around.
  • Record who proposed each rule. A MUST you added beyond what the user said is marked as your proposal in the amendment block until they confirm it. Never credit the user with your extension.
  • A gate that passes with nothing to check is hollow. "The test suite MUST be green" holds with zero tests. Name the behavior that must be proven, or require the suite to cover it.

Report

The version (old → new), the principles added, changed or removed, the documents rippled or still pending, and any TODO left open.

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.