agentleFS
Sign inStart free
backnotpropbacknotprop/product-engineering/.claude/skills/foundry/SKILL.md

foundry

Skill8 starsChanged 3 months agoApache-2.0View on GitHub

The build-and-maintenance process for this repository's converged skill kit. Use for ANY convergence or maintenance work here — lifting files from upstream sources, building or revising one of the six skills, handling upstream drift or a failed integrity check, adding a contradiction ruling, distilling source material, installing the private course drop-in, or evaluating a new upstream or skill for the kit. Load BEFORE editing anything.

5 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions5 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 5

    1. razakiau/skills★ 3same name
    2. huuphuoc1396/android-claude-skills★ 1same name
    3. LiorZ/foundry-claude-skill★ 1same name
    4. mjmirza/foundry★ 0same name
    5. ryktara/Foundry★ 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

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

  • 0

    Common

    Most of the 5 edited versions we read keep these.

  • 1

    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 foundry/MANIFEST.json — look up the class of every file you intend to touch.

    line 19
  2. Very specific:

    Read foundry/LEDGER.md — rulings bind everything you author.

    line 20
  3. Distinctive:

    Anti-slop gates on anything you author: no restatement between spine and references (principle in the spine, recipe in the ref, never both) …

    line 45
  4. Very specific:

    Plan first: write foundry/derivations/<skill>.md (sources, classes, cuts, rulings applied) before running a single lift.

    line 52

“Common” is measured: kept by at least 6 in 10 of the 5 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

Foundry

This repo converges 84 upstream design-engineering skills into six (pe-design, pe-build, pe-review, pe-verify, pe-product-description, pe-brand-assets), preserving the best source prose byte-for-byte and proving it with hashes. You are operating a factory with strict rules, not editing a normal repo. AGENTS.md is the fence; this skill is the manual.

Before anything else

  1. Read foundry/MANIFEST.json — look up the class of every file you intend to touch.
  2. Read foundry/LEDGER.md — rulings bind everything you author.
  3. Pick the runbook that matches the task (table below) and follow it exactly.
  4. End every session that changed provenance state with a foundry/LOG.md line and a green foundry/scripts/check-integrity.sh.

File classes

ClassMeaningYour permissions
verbatimByte-for-byte lift from a pinned upstream commitNever edit. Re-lift via script or nothing.
verbatim-minusVerbatim with a recorded cut (pristine + .patch in foundry/pristine/)Never edit the file directly. Changing the cut = edit file, run make-patch.sh, PR shows the patch diff.
distilledAuthored condensation of recorded sources; the salvage list in its derivation file is the contractEdit only via the distill-refresh runbook; finish with check-integrity.sh --update-distilled <path>.
authoredOurs: spines, glue, foundry docsEdit freely, subject to the ledger and the anti-slop gates.

private: true entries are course drop-in files — purchased content that exists only on local machines, never in a commit.

The convergence method (for building or revising a skill)

  • Architecture: each skill is a thin authored spine (router: modes, cues, contract, rulings block; target < 150 lines) over references lifted verbatim from graded sources. New prose is written only for glue, routing, and rulings.
  • Grades bind handling: GOLD → lift verbatim, never reword (cut whole sections only, via verbatim-minus, logged) · SOLID → lift, dedupe only · PADDED → distill against a named salvage list · SLOP → exclude.
  • Anti-slop gates on anything you author: no restatement between spine and references (principle in the spine, recipe in the ref, never both) · every sentence must change an implementation choice · no adjective stacks, personas, or dial theater · skill activation load stays under ~500 lines · no provenance in runtime text — MANIFEST, derivations, and NOTICE own provenance; skill prose states rules without citing sources, patches, pins, or ledger ids (frontmatter carries only a one-line derivations pointer).
  • Plan first: write foundry/derivations/<skill>.md (sources, classes, cuts, rulings applied) before running a single lift. The receipts precede the work.
  • Verify: check-integrity.sh green, skills-ref validate (or equivalent) clean, spine checked against every ledger ruling it touches.

Runbooks

TaskRunbook
Bring a file in from an upstreamrunbooks/lift.md
An upstream changed a vendored filerunbooks/re-lift.md
A distilled file's source changed, or the distillation needs workrunbooks/distill-refresh.md
Sources contradict, or a ruling needs to changerunbooks/new-ruling.md
CI integrity check is redrunbooks/integrity-failure.md
Install/refresh the purchased course files locallyrunbooks/course-dropin.md
Evaluate a new upstream or add a skill to the kitrunbooks/new-skill.md

Division of labor — never blur it

LLMs propose (grade, classify drift, author spines, draft PRs) · hashes verify (check-integrity.sh, in CI on every push) · humans merge (nothing you produce lands on main without review). Your token stream is never the channel vendored prose travels through.

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