adonis-skills
CLAUDE.md4 starsChanged 5 days agoMITView on GitHub
pnpm dev serves the web app at http://localhost:3000. After a UI change, open it and check a screenshot before reporting done.
In this repository43 other files it gives its agents.
- llms.txtllms.txt
- agent-symlink-initSkill
- code-inspector-initSkill
- commitSkill
- copywritingSkill
- css-tailwind-stylingSkill
- lingui-best-practicesSkill
- lingui-workflowSkill
- repo-skill-creatorSkill
- weekly-reportSkill
- agent-dispatchSkill
- agentic-review-handoffSkill
- agent-practice-auditSkill
- agent-symlink-initSkill
- architecture-hardening-loopSkill
- ardot-mcpSkill
- branch-creatorSkill
- chrome-dev-mcpSkill
- code-inspector-initSkill
- code-plugin-architectureSkill
- commit-pushSkill
- commitSkill
- decision-first-technical-writingSkill
- design-systemSkill
- desktop-agent-activitySkill
- dev-environment-roi-auditSkill
- disk-tidySkill
- figma-mcpSkill
- file-tidySkill
- goal-gateSkill
- grill-autopilotSkill
- kimi-computer-useSkill
- lingui-workflowSkill
- local-web-surfaceSkill
- open-code-review-loopSkill
- prompt-studioSkill
- review-prompt-composerSkill
- task-completion-loopSkill
- uxc-facadeSkill
- web-performance-auditSkill
- weekly-reportSkill
- workflow-gateSkill
- worktree-tidySkill
The file
Show its rules and text
We split this CLAUDE.md into the rules it gives an agent
19 rules. About 0 are common, 9 are distinctive, and 10 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈9
Distinctive · estimate
This author's own choices. Worth a look.
10
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- could be automatic: a setting could enforce it
- Distinctive: line 6
Use English for code, file names, commands, and commit messages.
- Distinctive: line 11
Before committing, inspect the branch, upstream, staged diff, and verification.
- Very specific: line 48
Use pnpm skills:new for the normal create, validate, and index flow.
- Very specific: line 49
Add OpenAI metadata only when needed, using pnpm skills:openai-yaml.
- Distinctive: line 54
Keep changes minimal and match existing script and file naming.
- Distinctive: line 55
Prefer deterministic scripts over repeated manual steps.
- Distinctive: line 62
Keep both files structurally and semantically aligned, including commands, and add reciprocal language links.
- Distinctive: line 63
Every README change must update both files in the same change.
- Very specific: line 67
Use lowercase hyphen-case for skill directories and frontmatter name.
- Very specific: line 68
Every skills/<slug>/SKILL.md requires non-empty name and description.
- Very specific: line 69
Add scripts/, references/, assets/, or other documentation only when operationally needed.
- Very specific: line 70
Add agents/openai.yaml only for downstream OpenAI metadata integration and generate it with pnpm skills:openai-yaml.
- Very specific: line 75
When you add a public skill or change what one does, write its apps/web/content/skill-catalog.yaml entry yourself in the same change.
- Distinctive: line 97
Use TypeScript and function components.
- Distinctive: line 98
Document non-obvious tradeoffs in PR descriptions.
- Very specific: line 99
Prefer ahooks useLocalStorageState over direct window.localStorage.
- Very specific: line 105
Create one docs/plan-YYYY-MM-DD-topic.md per major change.
- Very specific: line 110
Before changing web visuals, read the root DESIGN.md (what the site looks like) and docs/web-ui-clay-contract.md (how this repo implements …
- Distinctive: line 112
Do not apply its visual rules to scripts, skills, or non-web packages.
≈ “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
Repository Instructions
Language
- Respond in Chinese unless the user requests another language.
- Use English for code, file names, commands, and commit messages.
Git Delivery Authorization
- An explicit
commit-pushinvocation authorizes committing the selected change set and pushing the current branch to its configured upstream, includingmainormaster; do not ask for a second protected-branch confirmation. - Before committing, inspect the branch, upstream, staged diff, and verification. Never force-push. Stop for secrets, unrelated staged changes, a missing or unexpected push target, or a failed push.
Repository Model
skills/<slug>/contains public skills indexed by the web app;.agents/skills/contains internal tooling and is not published..claude/skillsis a symlink to.agents/skillsfor local runtime testing.skills/desktop-agent-activity,skills/installed-app-feature-inventory, andskills/dev-environment-roi-auditare exported from a private source (listed in.export-manifest.json). Treat them as generated: edit the private source, not these copies; the next export overwrites local edits.apps/web/src/generated/skills-index-lite.json,skills-detail-index.json, andskills-catalog-layout.jsonare generated web build inputs. After changing public skills, runpnpm skills:validateandpnpm skills:index; never edit the index manually.- Every public skill needs an entry under
skills:inapps/web/content/skill-catalog.yaml(Display Name, Scenario Group, Visitor-facing tags, summary, and use cases in English and Chinese); the same file holds the orderedgroupsandfeaturedlists.pnpm skills:validatefails without them. Keep this copy out of SKILL.md frontmatter: the Agent Skills spec allows only stringmetadatavalues.
Install Targets
| Host | Command |
Codex, Cursor, other skills CLI hosts | npx skills add adonis0123/adonis-skills --skill <name> (-g global, --list to list) |
| Claude Code | /plugin marketplace add Adonis0123/adonis-skills, then /plugin install adonis-skills@adonis-skills (manifest: .claude-plugin/marketplace.json) |
| Hermes Agent | hermes skills tap add adonis0123/adonis-skills |
Commands
pnpm skills:new
pnpm skills:init <skill-name> --path skills
pnpm skills:quick-validate skills/<skill-slug>
pnpm skills:openai-yaml skills/<skill-slug>
pnpm skills:validate
pnpm skills:index
pnpm skills:install:local [-- --all | --skill <name>]
pnpm skills:test:local [-- --all | --skill <name>]
pnpm dev
pnpm lint
pnpm typecheck
pnpm buildpnpm dev serves the web app at http://localhost:3000. After a UI change, open it and check a screenshot before reporting done.
- Use
pnpm skills:newfor the normal create, validate, and index flow. - Add OpenAI metadata only when needed, using
pnpm skills:openai-yaml. - For app/runtime changes, run the relevant lint, typecheck, and build checks.
Repository Conventions
- Keep changes minimal and match existing script and file naming.
- Prefer deterministic scripts over repeated manual steps.
- Change generated artifacts through their source script, then regenerate them.
- Pass script arguments explicitly through
--in docs and CI when pnpm parsing would be ambiguous.
Documentation Conventions
- A documented directory must contain canonical English
README.mdand matching ChineseREADME.zh-CN.md. - Keep both files structurally and semantically aligned, including commands, and add reciprocal language links.
- Every README change must update both files in the same change.
Skill Authoring
- Use lowercase hyphen-case for skill directories and frontmatter
name. - Every
skills/<slug>/SKILL.mdrequires non-emptynameanddescription; CI validates these two fields. - Add
scripts/,references/,assets/, or other documentation only when operationally needed. - Add
agents/openai.yamlonly for downstream OpenAI metadata integration and generate it withpnpm skills:openai-yaml. - Public skills live in
skills/; internal skills live in.agents/skills/and are not indexed.
Catalog entry for a new or changed public skill
When you add a public skill or change what one does, write its apps/web/content/skill-catalog.yaml entry yourself in the same change; do not leave it for the user. Follow the root DESIGN.md and the existing entries:
displayName: a short Title Case English name a Visitor understands without knowing the slug, plus a natural Chinese name (not a word-for-word translation). The slug stays the install identifier.group: the one existing Scenario Group a Visitor would look in. Add a new group only when none fits; a new group needstitle,summary(en/zh) and an illustration.summaryanduse_casesin English and Chinese, written for a Visitor, not for the agent;tagsfromSKILL_TAGS.featuredis curated by the maintainer: change it only when asked. A newly featured skill needs its own illustration.- Illustrations follow
DESIGN.md→ Illustration: transparent clay art generated with Codex image generation from the hero reference, no text. Add the PNG and WebP toadonis-brand/projects/skills/illustrations/, register the copy inadonis-brand/data/consumers.json, then copy the WebP toapps/web/public/brand/illustrations/. - Then run
pnpm skills:validateandpnpm skills:indexand stage the regeneratedapps/web/src/generated/. The pre-commit hook runsscripts/precommit-skills.mjs, which blocks a commit touchingskills/or the catalog when validation fails or the generated index is stale.
---
name: skill-name
description: "When and why to use this skill"
allowed-tools: Read, Bash # optional
metadata: # optional
author: your-name
version: "1.0.0"
---Web Coding (apps/web/**)
- Use TypeScript and function components; follow existing imports, naming, and nearby test placement.
- Document non-obvious tradeoffs in PR descriptions.
- Prefer ahooks
useLocalStorageStateover directwindow.localStorage. If direct access is required, document why in code.
Plan Mode Major Changes
- This section applies only in Plan mode.
- A change is major when it crosses modules/directories, changes public interfaces/architecture/core data flow, or is expected to touch at least 5 files.
- Create one
docs/plan-YYYY-MM-DD-topic.mdper major change; use lowercase kebab-case matching^plan-[0-9]{4}-[0-9]{2}-[0-9]{2}-[a-z0-9-]+\.md$. - Include background, goals, scope, solution, risks, and acceptance criteria.
Web UI Contract (apps/web/src/** only)
- Before changing web visuals, read the root
DESIGN.md(what the site looks like) anddocs/web-ui-clay-contract.md(how this repo implements it); both are mandatory.DESIGN.mdis generated bypnpm brand:syncfrom the privateadonis-brandrepo: never edit it here. - Colors come from the design system in the private
adonis-brandrepo (design-system/tokens.json;projects/skills/theme.jsonpicks the accent theme, currentlyleaf). Components use design-system roles only.apps/web/src/styles/brand-tokens.cssandtheme-roles.cssare generated: change the brand repo, then runpnpm brand:sync; CI runspnpm brand:check. - Do not apply its visual rules to scripts, skills, or non-web packages.
- In particular, never add bottom outer highlight lines such as
0 1px 0 ...to raised or floating shadows.
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.

