agentleFS
wyattjohwyattjoh/skills/CLAUDE.md

skills

CLAUDE.md3 starsChanged 46 days agoMITView on GitHub

This repository is a public collection of Claude Code skills and agents for software development workflows. There is no application to build: skills are Markdown (SKILL.md) with optional Bun/TypeScript helper scripts, and agents are single Markdown files. For how Claude Code auto-loads skills, agents, rules, and memory, see .claude/references/claude-code-loading.md. Skills under skills/ are the published collection and belong in the README.md table. Skills under .claude/skills/ are repo-internal tooling: they load automatically when working in this repository and are not listed…

The file

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

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

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    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:

    Follow the frontmatter conventions in .claude/rules/agents.md.

    line 57
  2. Very specific:

    Update README.md whenever a skill or agent is added, removed, or renamed, so the agents and skills tables match the actual contents of …

    line 83
  3. Very specific:

    Use Conventional Commits: <type>[scope]: <description>.

    line 125
  4. Distinctive:

    Never use em dashes in written output, including comments, commit messages, and docs.

    line 126

≈ “Common” is an estimate: at least 1 in 5 of the 26,980 CLAUDE.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
CLAUDE.md

CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

Overview

This repository is a public collection of Claude Code skills and agents for software development workflows. There is no application to build: skills are Markdown (SKILL.md) with optional Bun/TypeScript helper scripts, and agents are single Markdown files.

For how Claude Code auto-loads skills, agents, rules, and memory, see .claude/references/claude-code-loading.md.

Repository structure

skills/              # one directory per published skill, each with a SKILL.md (plus optional scripts/ and references/)
agents/              # one Markdown file per agent
plugins/             # optional plugins beside the root one (e.g. to-code: a Claude Code mod plus a pi extension)
.claude-plugin/      # plugin.json and marketplace.json, so the repo installs as a Claude Code plugin
.claude/skills/      # skills internal to this repo, not published (e.g. writing-skills-update)
.claude/references/  # shared reference docs for authoring in this repo, plus pinned dependency submodules
.claude/rules/       # path-scoped authoring conventions for this repo
package.json         # Bun workspace root (workspaces: ["skills/*", "plugins/*"])
justfile             # symlink management for installing agents/ into ~/.claude/agents/

Skills under skills/ are the published collection and belong in the README.md table. Skills under .claude/skills/ are repo-internal tooling: they load automatically when working in this repository and are not listed in README.md.

Path-scoped rules

Conformance rules live in .claude/rules/ and auto-activate via paths: frontmatter when Claude edits matching files:

RuleScopePurpose
skills.mdskills/**SKILL.md format, Bun/TypeScript scripts, directory structure
agents.mdagents/**/*.mdAgent frontmatter fields, reload behavior, validation
plugins.mdplugins/**Shared mod/pi core, adapters, the two test runners
testing.mdAll filesPositive, exact test expectations instead of negated chains

Authoring skills

  • Invoke the skills-skill skill before creating or editing a skill.
  • One directory per skill under skills/, containing a SKILL.md. Co-locate scripts/ and references/ inside the skill directory as needed.
  • Frontmatter requires name and a description with trigger phrases so the skill activates reliably.
  • Helper scripts use Bun + TypeScript. Run bun test to exercise their tests.

Authoring agents

  • One Markdown file per agent under agents/. Restart Claude Code to load agent changes.
  • Follow the frontmatter conventions in .claude/rules/agents.md.

Plugin packaging

The repository root is a Claude Code plugin (wyattjoh) and the marketplace (wyattjoh-skills), defined in .claude-plugin/. The marketplace also lists each optional plugin under plugins/, so installing wyattjoh never pulls them in. The root plugin relies on the default skills/ and agents/ directories, so new skills and agents ship without manifest edits. version is intentionally omitted so installs track the commit SHA. Run claude plugin validate . after touching either manifest.

Available scripts

CommandDescription
bun testRun all tests in the skills/ directory
bun run checkType-check skill TypeScript files (tsc --noEmit)
bun run lintLint with oxlint (config: .oxlintrc.json)
bun run formatFormat with oxfmt (config: .oxfmtrc.json)
bun run format:checkVerify formatting without writing
bun run knipFind unused files, deps, and exports (knip.json)
just linkSymlink agents/*.md into ~/.claude/agents/
just unlinkRemove agent symlinks pointing at this checkout

Documentation

Update README.md whenever a skill or agent is added, removed, or renamed, so the agents and skills tables match the actual contents of agents/ and skills/.

Dependency References

Upstream repositories are registered as pinned git submodules under .claude/references/. They are for read-only reference only; do not edit files inside these paths.

DependencyVersion / TagPathRepositoryPin (commit SHA)
Catppuccinv0.2.0.claude/references/catppuccinhttps://github.com/catppuccin/catppuccin.git9de299f8f1702fe4fb4e439adfd04b5623e7b77f
Composable Architecture1.26.2.claude/references/swift-composable-architecturehttps://github.com/pointfreeco/swift-composable-architecture.git377da4061db10d26337a71bb279c506bb951f50f
Effecteffect@3.22.2.claude/references/effecthttps://github.com/Effect-TS/effect.git6985be0cf461f0997f28f6798f469d01a2b46ca3
Effect v4effect@4.0.0-rc.118.claude/references/effect-v4https://github.com/Effect-TS/effect.gitad61db80efd52637e2901c5ee56b9e0fb4e8ac48
jev-doc-searchmain@7b1008c.claude/references/jev-doc-searchhttps://github.com/VectifyAI/jev-doc-search.git7b1008c437c8c9a88da8a427a76052fb0454977a
Varlockvarlock@1.21.1.claude/references/varlockhttps://github.com/dmno-dev/varlock.gitfd51d60905b372ab10ae6407300346e67695aec0

The Effect repository is vendored twice because this repository uses two incompatible major versions. The Effect references are documentation snapshots. effect supplies v3 source for the unchanged workspaces, clean-storage, linked-docs, mermaid, and reference-submodules scripts; effect-v4 tracks the exact v4 release candidate documented by the effect-ts skill. Migrated runtime scripts target the current Alchemy-compatible stable release separately from that documentation corpus. For script API changes, use the skill's package manifest and matching installed Effect source. The v4 checkout also carries upstream MIGRATION.md, migration/, LLMS.md, and ai-docs/ directories that the v3 checkout does not.

The alchemy and effect-ts skills must always target the same Effect release. The source of truth is upstream alchemy's exact pin (overrides.effect in its pnpm-workspace.yaml), which skills/alchemy/scripts/build-skill.ts records as source.effectVersion in skills/alchemy/references/manifest.json. skills/alchemy/scripts/effect-version.test.ts fails unless every Effect v4 version in skills/effect-ts/ and the effect-v4 submodule tag match it, so after regenerating the alchemy corpus, bump the effect-ts skill and the effect-v4 submodule together in the same change.

To populate locally after a fresh clone:

git submodule sync
git submodule update --init --depth 1

Conventions

  • Use Conventional Commits: <type>[scope]: <description>.
  • Never use em dashes in written output, including comments, commit messages, and docs. Use commas, parentheses, or separate sentences instead.

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.