AGENT_RESOURCE
Health Gecti
- License — License: Apache-2.0
- Description — Repository has a description
- Active repo — Last push 0 days ago
- Community trust — 78 GitHub stars
Code Basarisiz
- exec() — Shell command execution in skills/standalone/deepseek-mcp/deepseek_mcp.mjs
- process.env — Environment variable access in skills/standalone/deepseek-mcp/deepseek_mcp.mjs
- network request — Outbound network request in skills/standalone/deepseek-mcp/deepseek_mcp.mjs
- process.env — Environment variable access in skills/vipin/communication-assistant/scripts/outbox.js
- network request — Outbound network request in skills/vipin/email-assistant/pnpm-lock.yaml
Permissions Gecti
- Permissions — No dangerous permissions requested
Bu listing icin henuz AI raporu yok.
My own agent skills for Claude Code and Codex, plus pointers to the third-party skills I use.
AGENT_RESOURCE
My own agent skills for Claude Code and Codex, plus pointers to the third-party skills I use.
I use Claude Code and Codex on a Windows machine, and these are the skills I wrote for
my own chores: keeping my drives organized, drafting replies across chat apps and Gmail,
turning books into manga scripts, publishing videos, and making an agent understand a
project before it edits anything. Skills written by other people are not copied here any
more; THIRD_PARTY_SKILLS.md lists them with pinned upstream
commits, and a script downloads them for you.
$ python tools/skill_catalog.py validate
Validated with 0 error(s) and 0 warning(s).
$ python tools/skill_catalog.py search "turn a novel into a manga script" --limit 1
33 literature-to-manga skills/vipin/literature-to-manga/SKILL.md
Convert selected prose, novel, biography, history, or text-archive works into manga adaptation packages. Use when ask...
$ scripts/fetch-third-party.sh superpowers gsap-skills
superpowers -> third_party/superpowers @ f2cbfbe
gsap-skills -> third_party/gsap-skills @ aed9cfd
My skills
| Skill | What it does | Needs |
|---|---|---|
| workstation-maintenance | Inventories my drives, plans file moves in approval batches, moves only approved batches, and can roll them back | Windows, PowerShell 5.1+ |
| project-hardness | Has the agent map a project (architecture, constraints, blast radius) and turn a vague request into a task spec before editing | Python 3 and my WEIPING_WIKI repo, which holds the engine |
| literature-to-manga | Turns a novel or biography into a manga adaptation package: brief, bible, panel scripts, image prompts | Python 3.10+ |
| communication-assistant | Triage and draft replies for WeChat, WhatsApp, QQ, Feishu and email; keeps a local outbox and never sends on its own | Node.js 18+ |
| email-assistant | Gmail API scripts to search, read, summarize and draft mail; sending needs an explicit confirm flag | Node.js 18+, a Google Cloud OAuth client |
| feishu-bridge | Routes Feishu/Lark tasks to lark-cli and the official lark-* skills |
lark-cli and the Lark skills |
| wechat-video-channel-publish | Playwright CLI that logs into WeChat Channels (视频号) and uploads, schedules or drafts a video | Node.js 20+ |
| lidang-perspective | Persona skill built from Li Dang's public writing, for career and tech-path questions (in Chinese) | nothing |
| deepseek-mcp | Tiny MCP server that lets Codex call DeepSeek as a side tool | Node.js 18+ or Python 3, a DeepSeek API key |
Each folder has a README with one usage example. SKILL-INDEX.md has
the "use this when..." line for each skill, which is what I point my agents at.
Nothing here hard-codes my own folders. Machine-specific settings come from a
git-ignored config file (workstation-maintenance/config.json, copied fromconfig.example.json) or from environment variables that each skill's README names
(EMAIL_ASSISTANT_AUTH_DIR, COMMUNICATION_ASSISTANT_HOME, LARK_CLI,LITERATURE_BOOKS_DIR).
Install a skill
A skill is just a folder with a SKILL.md. Clone the repo and link the folder into your
agent's skills directory (~/.claude/skills/ for Claude Code, ~/.codex/skills/ for Codex):
git clone https://github.com/appleweiping/AGENT_RESOURCE.git
New-Item -ItemType Junction -Path "$HOME\.claude\skills\literature-to-manga" `
-Target "$PWD\AGENT_RESOURCE\skills\vipin\literature-to-manga"
git clone https://github.com/appleweiping/AGENT_RESOURCE.git
ln -s "$PWD/AGENT_RESOURCE/skills/vipin/literature-to-manga" ~/.claude/skills/literature-to-manga
Then start a new session and ask for the task in plain words; the agent picks the skill
from its description. Skills with scripts list their own install step (npm install,pip install pypdf, ...) in their README.
Third-party skills
scripts/fetch-third-party.sh # all of them, into ./third_party (git-ignored)
.\scripts\fetch-third-party.ps1 shadcn # Windows, just one
The list, the pinned commits and why each one is pinned to that commit are in
THIRD_PARTY_SKILLS.md.
Checks
python tools/skill_catalog.py validate --strict # SKILL.md frontmatter rules
python -m unittest discover -s tools/tests -p "test_*.py"
python tools/check_links.py # relative Markdown links resolve
GitHub Actions runs these plus a secret scan (tools/Test-PublicSafety.ps1) on every push.
License
Apache-2.0 (LICENSE) covers my own skills and scripts in this repo only.
Third-party skills keep their upstream licenses and are not redistributed here; see
THIRD_PARTY_SKILLS.md. One generated file insidewechat-video-channel-publish (src/stealth.min.js) comes from puppeteer-extra under
MIT; its copyright and license text are in that skill's
THIRD_PARTY_NOTICES.md.
Yorumlar (0)
Yorum birakmak icin giris yap.
Yorum birakSonuc bulunamadi