agentleFS
dmitribogatenkovdmitribogatenkov/briefproof/skills/change/SKILL.md

change

Skill47 starsChanged 5 days agoMITView on GitHub

Changes a project that Briefproof already built, with the same discipline at small scale. The request becomes new quoted requirements, a spec delta, only the tickets that change, and a blind re-check of the request plus a regression pass over what the base run delivered. Starts only on /briefproof:change.

9 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

3 rules. 0 are common, 0 are distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 8 edited versions we read keep these.

  • 0

    Distinctive

    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
  1. Very specific:

    Read {base}/4-spec/spec.md, {base}/4-spec/interfaces.md and the project's AGENTS.md before anything else.

    line 38
  2. Very specific:

    Briefing: Ask only what the request leaves open, one round, suggested answer first, through AskUserQuestion.

    line 51
  3. Very specific:

    Spec: Write {run}/4-spec/spec.md as a delta: start with one line naming the base spec, then only the sections that change, each saying what …

    line 53

“Common” is measured: kept by at least 6 in 10 of the 8 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

Briefproof · change request

First, check you are in Claude Code. This skill needs the plugin's ledger tool, mcp__briefproof__state. If it is not available, tell the user in one line, "Briefproof runs in Claude Code: open the Code tab (local session) or a terminal and run /briefproof:change", and stop.

A change request is a short Briefproof run on top of a finished one. Everything in /briefproof:build still holds (read its five rules if you have not this session): the request is quoted, never paraphrased; a stranger checks the result against the words; no code before the spec delta; one fresh executor per ticket. What differs is scope: you touch only what the request needs, and you prove the rest still works.

Start

state {action: "init", kind: "change", request: "<the user's words, verbatim>", mode: "ask"}

The ledger refuses when this folder has no finished run (tell the user to run /briefproof:build first) or when the run is unfinished (resume it instead). On success it creates a new run folder, {run}, and names the base run, {base}. The mod has already written:

  • {run}/1-brief/brief.md: the base brief with a ## Change request section holding the request verbatim. This is the brief for this run.
  • {run}/1-brief/base-requirements.md: what the base run delivered, for the regression pass.

Read {base}/4-spec/spec.md, {base}/4-spec/interfaces.md and the project's AGENTS.md before anything else. You are changing a known system, not discovering one.

Stages

The eight stages are the same and the pane shows them the same way; most are shorter.

  1. Preflight. Confirm the working tree is clean (commit or ask). stage manifest.
  2. Manifest. add_requirements with verbatim quotes from the change request section only. The base requirements are not re-added; they are the regression list.
  3. Briefing. Ask only what the request leaves open, one round, suggested answer first, through AskUserQuestion. Often nothing.
  4. Spec. Write {run}/4-spec/spec.md as a delta: start with one line naming the base spec, then only the sections that change, each saying what it replaces. Copy {base}/4-spec/interfaces.md to {run}/4-spec/interfaces.md and amend it; never edit the base run's files. Gate G2: the coverage checker reads {run}/1-brief/brief.md and the delta; tell it to check the change request section against the delta.
  5. Plan. Tickets for the change only, each naming the files it owns. If a change touches a shared boundary, the three-ticket rule from the build skill applies (add beside, migrate, remove). Gate G3, commit the plan.
  6. Build. Waves, fresh executors, reviews where it pays, exactly as in the build skill. Executors read {run}/4-spec/interfaces.md.
  7. Acceptance. Clone to a temp dir and spawn the blind checker with prompts/acceptance.md, giving it {run}/1-brief/brief.md, {run}/1-brief/base-requirements.md and the clone. Tell it: check every sentence of the Change request section as the brief; then, for each line of base-requirements.md, confirm it still holds with one quick check each and report any that broke. A broken base requirement is a fix ticket before anything else. Save the reply as {run}/7-acceptance/acceptance.md. gate G4.
  8. Report. {run}/8-report/report.md: what changed and why, the new R## table, the regression result (N of N base requirements still hold), anything deferred, and the usage table from show. Update AGENTS.md if how to run or test changed. stage done.

Red flags

  • You edited a file under {base}/. Base runs are history.
  • A base requirement broke and the report says "done" anyway.
  • You rewrote the whole spec instead of a delta: the reader loses what changed.
  • The request grew into a redesign. Say so, and offer /briefproof:build with a new brief instead.

Resume

A bare /briefproof:change with an unfinished change run in this folder: show, re-read {run}/1-brief/brief.md and the current stage's files, continue.

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.