agentleFS
Sign inStart free
janek-moonjanek-moon/murex/skills/ratchet/SKILL.md

ratchet

Skill0 starsChanged 3 months agoMITView on GitHub

Use when requirements are clear and you want to build a feature the disciplined way - bottom-up, verifying each layer before the next. Decompose into components with verifiable requirements, build the lowest one whose dependencies are all verified (in a fresh subagent), verify it with evidence, and ratchet upward until complete. The binary refuses to build on unverified ground. For work with real unknowns or risk, use spiral instead.

  • Pipes a download into a shell

4 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions4 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 4

    1. nickwinder/skills★ 1same name
    2. OmkarKamble1/ratchet★ 1same name
    3. SPixs/claude-skill-ratchet★ 1same name
    4. tamesystems/intent-engineering-skills★ 0same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository2 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

3 rules. 0 are common, 2 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 3 edited versions we read keep these.

  • 2

    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. Distinctive:

    Use this when the requirements are clear.

    line 19
  2. Distinctive:

    When it returns, verify the evidence yourself - run the checks - before touching the gate.

    line 75
  3. Very specific:

    If the build did not meet its requirement, murex ratchet rework C1 --note "<what failed>" returns it to the frontier for another attempt.

    line 82

“Common” is measured: kept by at least 6 in 10 of the 3 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

murex - Ratchet: bottom-up verified construction

Description

You conduct a ratchet: a feature with clear requirements, built from the bottom up, one verified layer at a time. Three parts, three jobs (mirror of the spiral):

  • murex ratchet - deterministic bookkeeping: the component register, the buildable frontier, the verification gate. It never executes work.
  • An executor - a fresh subagent: one component's requirement in, evidence out.
  • You - decompose the feature into components with verifiable requirements, hand each build to the executor, verify its evidence yourself, close the gate.

Use this when the requirements are clear. The reason it is more than "just build it bottom-up": the binary refuses to hand you a component until every dependency is verified, and refuses to record a verification without evidence - so you cannot build on unverified ground, and a level, once verified, is locked. For work with real unknowns, use /murex:spiral instead; a drained spiral hands off here.

Prerequisites

murex on PATH, new enough to have the ratchet. Feature-detect and (re)install if missing or stale:

murex ratchet --help >/dev/null 2>&1 || curl -fsSL https://raw.githubusercontent.com/janek-moon/murex/main/install.sh | sh

The loop

start -> add* (decompose) -> [ next -> build (subagent) -> verify ]* -> complete

1. Open the ratchet

murex --root . ratchet start "ship CSV export" --requirement "a user downloads a valid CSV of their data"

2. Decompose into verifiable components (leaves first)

Register the lowest parts first; a component may only depend on ids that already exist. Each --requirement must be checkable.

murex --root . ratchet add "CSV row encoder" --requirement "encodes one record to an RFC-4180 line, quotes embedded commas/quotes"
murex --root . ratchet add "CSV writer" --requirement "streams all rows with a header" --depends-on C1
murex --root . ratchet add "export endpoint" --requirement "GET /export returns text/csv for the current user" --depends-on C2

3. Take the next buildable component

murex --root . ratchet next

Returns the build brief for the lowest buildable component - its requirement and the evidence/interfaces of its already-verified dependencies.

4. Build it in a fresh subagent

Dispatch a fresh subagent with the brief and nothing else: it builds the smallest implementation that satisfies the requirement and reports its evidence (tests run, output shown). Your context stays on the register.

5. Verify, then gate

When it returns, verify the evidence yourself - run the checks - before touching the gate.

murex --root . ratchet verify C1 --evidence "cargo test csv_encoder green; RFC-4180 sample matches byte-for-byte" --cost 1.0

If the build did not meet its requirement, murex ratchet rework C1 --note "<what failed>" returns it to the frontier for another attempt.

6. Repeat until complete

Go back to step 3. Check progress with murex --root . ratchet status (verified/total, the current frontier). The ratchet is complete when every component is verified; confirm the whole against the start requirement.

Notes

  • State lives in .murex/ratchet.json; commit it to share the build ledger. It is independent of a spiral's .murex/spiral.json - a repo can run either.
  • A verified component is final - the ratchet does not slip. Decompose so that "verified" is a claim you can stand behind.
  • murex ratchet list groups components by state; murex ratchet stop --reason "..." abandons the ratchet.

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