laoitdev-skills
CLAUDE.md0 starsChanged 4 months agoMITView on GitHub
A curated collection of Agent Skills for Claude Code: self-contained playbooks that teach an agent project-specific conventions and workflows for other codebases (React/TypeScript frontends, Flutter mobile apps, etc.). This repo does not contain application code — it contains the skills themselves, distributed two ways: Both distribution paths must keep working, so don't restructure skills/ without checking both plugin.json's skills field and each category README.md. CI (.github/workflows/validate-skills.yml) runs validate-skills.sh on every push to main and every PR — it's the only…
In this repository12 other files it gives its agents.
- feature-architectureSkill
- frontend-auditSkill
- frontend-i18nSkill
- mui-themeSkill
- react-coreSkill
- react-standardSkill
- table-mrtSkill
- tanstack-formSkill
- tanstack-querySkill
- tanstack-routerSkill
- flutter-clean-arch-getxSkill
- flutter-flavorsSkill
The file
We split this CLAUDE.md into the rules it gives an agent
2 rules. About 0 are common, 1 is distinctive, and 1 is 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.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 51
Write the body as instructions to an agent, not documentation for a human — imperative, concrete, copy-paste-able patterns over abstract …
- Very specific: line 52
After adding/editing a skill, update that category's skills/<category>/README.md table.
≈ “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.
Files
CLAUDE.md
This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
What this repo is
A curated collection of Agent Skills for Claude Code: self-contained playbooks that teach an agent project-specific conventions and workflows for other codebases (React/TypeScript frontends, Flutter mobile apps, etc.). This repo does not contain application code — it contains the skills themselves, distributed two ways:
- Plugin/marketplace: the whole repo is a Claude Code plugin marketplace (
.claude-plugin/marketplace.json) bundling every skill into one plugin (.claude-plugin/plugin.json, namelaoitdev-skills). Users run/plugin marketplace add LaoitdevOpen/laoitdev-skillsthen/plugin install laoitdev-skills@laoitdev-skills. - Copy-paste: users copy an individual skill folder into their own project's
.claude/skills/or~/.claude/skills/.
Both distribution paths must keep working, so don't restructure skills/ without checking both plugin.json's skills field and each category README.md.
Commands
./scripts/validate-skills.sh # validate every skills/**/SKILL.md frontmatter (name, description present)
claude plugin validate . # validate .claude-plugin/marketplace.json and plugin.json schemaCI (.github/workflows/validate-skills.yml) runs validate-skills.sh on every push to main and every PR — it's the only automated check, so a passing local run means CI will pass.
To test the plugin end-to-end locally without publishing:
claude plugin marketplace add ./
claude plugin install laoitdev-skills@laoitdev-skills
# ... verify, then:
claude plugin uninstall laoitdev-skills@laoitdev-skills
claude plugin marketplace remove laoitdev-skillsRepository structure
skills/
├── backend/ # empty — contributions welcome
├── frontend/ # 9 skills — React + TypeScript conventions
└── mobile/ # 2 skills — Flutter conventionsEach skill is skills/<category>/<skill-name>/SKILL.md, optionally with references/ (extra docs loaded on demand) and templates/ (files the skill copies/adapts). .claude-plugin/plugin.json's skills field lists all three category directories explicitly (./skills/frontend, ./skills/backend, ./skills/mobile) — this is required because Claude Code's default plugin scan only looks one level under skills/ for <name>/SKILL.md, and this repo nests an extra category level.
Adding or editing a skill
Full rules are in CONTRIBUTING.md; the load-bearing ones:
- Folder:
skills/<category>/<skill-name>/. Categories today:backend,frontend,mobile. - Frontmatter
namemust exactly match the<skill-name>folder name (kebab-case) —validate-skills.shenforces this. - Frontmatter
descriptionis the trigger mechanism: an agent matches it against the task to decide whether to invoke the skill. State when to use it (concrete trigger phrases a user might say), not just what it does — see any existingSKILL.mdfor the density expected (e.g.skills/frontend/frontend-audit/SKILL.md). - Write the body as instructions to an agent, not documentation for a human — imperative, concrete, copy-paste-able patterns over abstract descriptions.
- After adding/editing a skill, update that category's
skills/<category>/README.mdtable. - Frontend skills follow a shared convention of a
## Core Invariants (always enforced — never violate)section near the top, listing the small set of rules that make the skill's target codebase pattern non-negotiable (e.g.tanstack-router's "route files live only insrc/routes/"). Follow this pattern for new skills that have hard invariants rather than burying them in prose.
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
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.

