LoopyLuci/Skills

最近记录的来源活动: 2026年10月13日 23:28

SkillsMP 收录数据更新: 2026年11月12日 12:52

已收集 skills: 10,021

GitHub 星标: 1
GitHub Forks: 1

用 AI 助手安装

复制这段 Prompt，发给你正在使用的 AI 助手。

Follow https://skillsmp.com/skill-install/prompt.md to install Agent Skills from https://github.com/LoopyLuci/Skills.

仓库 README

仓库 README 来源：README.md

查看 README 源文件

展开完整 README

# LoopyLuci Skills A curated collection of **10,107 AI agent skills** built for [Hermes Agent](https://hermes-agent.nousresearch.com/) and compatible with any agent that supports the [agentskills.io](https://agentskills.io/) open standard. Skills cover every major domain — software development, data/AI, cybersecurity, cloud, business, science, healthcare, engineering, creative, education, government, and more — across **5,789 distinct tags**. The repository holds **12,800+ tracked files** (~60 MB of skill content) and loads skills on demand. > Counts on this page are read from the repository. The collection grows as skills > sync in from connected Hermes instances; regenerate the per-skill index with > `python -m ci --only build && python scripts/generate-docs-index.py`. --- ## Quick Start ```bash # Clone the repo git clone https://github.com/LoopyLuci/Skills.git ~/LoopyLuci-skills cd ~/LoopyLuci-skills # Option A — external skills directory (recommended; read-only scan) # Add to ~/.hermes/config.yaml: # skills: # external_dirs: # - ~/LoopyLuci-skills/skills # Option B — skill tap (hub-managed updates) hermes skills tap add LoopyLuci/Skills hermes skills tap list # Option C — install a single skill by URL hermes skills install https://raw.githubusercontent.com/LoopyLuci/Skills/main/skills/<skill-name>/SKILL.md # Option D — inspect installed skills and browse what is available hermes skills list hermes skills browse ``` No per-skill install is required for Options A and B: every skill is available immediately as a slash command. --- ## Repository Layout ``` . ├── skills/ # 10,107 skill directories (flat) │ └── <skill-name>/ │ ├── SKILL.md # required — the skill itself │ ├── references/ # optional — supporting docs │ ├── scripts/ # optional — helper scripts │ ├── templates/ # optional — output templates │ ├── assets/ # optional — images, data │ ├── agents/ # optional — subagent definitions │ └── examples/ # optional — worked examples ├── scripts/ # tooling: generator, import, sync, docs │ ├── skill-generator/ # skill creation/validation/enrichment │ ├── import-upstream-skills.py# import skills from upstream repos │ ├── prep-customization.py # build the customization work list │ ├── assemble-custom-skills.py# assemble rewritten skills │ ├── generate-docs-index.py # regenerate docs/skills-index.md │ └── sync-*.py|.sh # Hermes ↔ repo synchronisation ├── ci/ # local on-device CI/CD pipeline │ ├── stages/ # preflight, static, skills, tests, build, smoke, deploy │ ├── hooks/ # git pre-commit / pre-push │ ├── config.json # thresholds, allowlists, patterns │ └── README.md # pipeline documentation ├── tests/ # stdlib unittest suite for the pipeline ├── docs/ │ └── skills-index.md # every skill, grouped by category ├── _sources/ # upstream clones used by the importer (git-ignored) └── dist/ # index + release archive (git-ignored) ``` --- ## Finding a Skill 1. **Complete index** — [docs/skills-index.md](docs/skills-index.md) lists all 10,107 skills with descriptions, grouped by category and linked to each `SKILL.md`. 2. **Machine-readable** — `dist/skills-index.json` (built by `python -m ci --only build`) carries name, title, description, tags, license, source, and file/byte counts. 3. **By tag** — search frontmatter `metadata.hermes.tags`, e.g. `grep -rl "tags:.*kubernetes" skills --include=SKILL.md`. 4. **By name** — skills are flat: `skills/<name>/SKILL.md`. 5. **Installed view** — `hermes skills list` and `hermes skills browse`. --- ## Skill Format Specification ### Directory structure Each skill is a directory named exactly like its frontmatter `name`: ``` skills/<skill-name>/ ├── SKILL.md # required ├── references/ # optional ├── scripts/ # optional ├── templates/ # optional ├── assets/ # optional ├── agents/ # optional └── examples/ # optional ``` ### `SKILL.md` frontmatter ```yaml --- name: skill-name # required — kebab-case, must match the directory description: Use when ... # required — one sentence, ideally <= 120 chars version: 1.0.0 # required — semver author: LoopyLuci Community # required license: MIT # required — SPDX identifier (see License below) platforms: [any] # required — [linux, macos, windows, any] metadata: hermes: tags: [tag1, tag2] # required — 2-5 lowercase tags recommended source: owner/repo # required for adapted skills (attribution) adapted: true # marks a skill rewritten from an upstream source --- ``` ### Body structure ```markdown # Skill Title One-sentence summary of what the skill does. > Adapted notice (adapted skills only) ## Trigger - Conditions that activate this skill ## When NOT to Use - Conditions that belong to another skill ## Core Sections... ### ... ## Related Skills in This Repo - [other-skill](../other-skill/SKILL.md) — description ## Adaptation Notes # adapted skills only - ... ## Source & Attribution # adapted skills only - Upstream, path, license, retrieval date ``` ### Tag guidelines - Lowercase, kebab-case for multi-word tags: `machine-learning`, `data-engineering`. - 2–5 tags is the norm (the repo median is 3; the CI check requires at least one). - Be specific: `postgresql` over `database`, `react` over `frontend`. - Reuse existing tags for consistency — the CI pipeline flags self-referential tags. --- ## Categories Counts below come from [docs/skills-index.md](docs/skills-index.md). | Category | Skills | | --- | --- | | Software Development & Engineering | 2,603 | | Science, Research & Healthcare | 1,639 | | Business, Product & Management | 1,536 | | General & Other | 1,441 | | Data Science, AI & Machine Learning | 852 | | Cloud & Infrastructure | 596 | | Engineering & Manufacturing | 597 | | Creative, Design & Media | 426 | | Cybersecurity | 264 | | Education & Personal Development | 129 | | Government, Law & Public Policy | 24 | | **Total** | **10,107** | The most common tags are `biotech` (574), `business` (443), `data` (388), `email` (388), `management` (355), `backend` (350), `cloud` (320), `engineering` (318), `security` (315), and `finance` (310). --- ## Local CI/CD Pipeline A fully custom, on-device pipeline — **stdlib-only, no third-party Python packages**. See [ci/README.md](ci/README.md) for the full reference. ```bash python -m ci --full # every stage python -m ci --fast # preflight, static, skills, tests python -m ci --only skills tests python -m ci --list # show the plan python -m ci --full --push ``` | Stage | What it does | | --- | --- | | `preflight` | Python version, git branch/commit/dirty state, remote, free disk, tool availability, config sanity | | `static` | `compileall`; `ruff`; secret, bare-`except`, oversize-file and JSON scans (working tree only) | | `skills` | Parallel validation of every `skills/*/SKILL.md` — frontmatter, name/dir match, license, tags, attribution, cross-links, support files, duplicates, upstream overlap | | `tests` | `unittest` discovery with hard timeout and flaky rerun | | `build` | `dist/skills-index.json` + release archive | | `smoke` | Artifact integrity: index parses, counts match, samples parse, archive verified | | `deploy` | Remote reachability + receipt; pushes only with `--push` on a clean tree | Typical runtime on this repository: **~21 s full**, **~9.5 s fast** across 10,107 skills and ~12,800 working-tree files. Safety features: PID lock with stale takeover, per-command timeouts with process-tree kill, change-aware stage selection, JSON reports with history and pruning, and a flaky-test rerun that separates flakes from real failures. ### Automation ```bash python -m ci.install_hooks # pre-commit (preflight+static) + pre-push (fast) python -m ci.watch # poll watcher; runs --fast on change python -m ci.watch --once # single run, for schedulers ``` Scheduled jobs (Hermes cron, delivering locally): - `skills-sync-hermes-repo` — every 30 m, Hermes ↔ repo sync. - `skills-ci-fast` — every 60 m, fast pipeline. - `skills-ci-full-daily` — daily at 03:00, full pipeline including build/archive. --- ## Upstream Import & Customization 59 skills are adapted from six upstream community repositories rather than copied. The pipeline lives in `scripts/`; full details in [docs/IMPORTING.md](docs/IMPORTING.md). | Upstream | Skills | | --- | --- | | [addyosmani/agent-skills](https://github.com/addyosmani/agent-skills) | 23 | | [oracle/skills](https://github.com/oracle/skills) | 19 | | [mattpocock/skills](https://github.com/mattpocock/skills) | 7 new (34 already present) | | [obsidianmd/obsidian-skills](https://github.com/obsidianmd/obsidian-skills) | 7 | | [cloudflare/security-audit-skill](https://github.com/cloudflare/security-audit-skill) | 1 | | [Forward-Future/loopy](https://github.com/Forward-Future/loopy) | 2 | Each adapted skill carries a rewritten description, derived Trigger and When-NOT-to-Use sections, a condensed/restructured body, real cross-links into this repo, adaptation notes, and license-correct attribution. The CI pipeline flags any adapted skill that drifts back toward verbatim upstream text. --- ## Automatic Sync The repository syncs bidirectionally with Hermes Agent instances on a schedule (currently every 30 minutes): ```bash ./scripts/sync-from-hermes.sh # dry run ./scripts/sync-from-hermes.sh --apply # copy changes in ./scripts/sync-from-hermes.sh --force # sync, commit, and push ``` Other sync helpers: `scripts/sync-bidirectional.py`, `scripts/sync-skills-fast.py`, `scripts/sync-hermes-windows.py`, `scripts/sync-machine.py`. --- ## Tests `tests/` contains the stdlib `unittest` suite for the pipeline itself — validation rules and orchestration units. It runs as the `tests` stage: ```bash python -m unittest discover -s tests -t . -v ``` 27 tests cover frontmatter parsing, every skill rule, planning, change selection, locking, pruning, and the secret-scanning heuristics. --- ## Skill Generator Toolkit ```bash # Generate one skill python scripts/skill-generator/generate_skill.py generate \ --name my-skill --description "Does X" --tags python,cli [--category CAT] --apply # Bulk generate from a JSON file ([{"name","description","tags"}, ...]) python scripts/skill-generator/generate_skill.py bulk skills.json --apply # Validate one skill python scripts/skill-generator/generate_skill.py validate skills/my-skill # Enrich boilerplate skills with real content python scripts/skill-generator/generate_skill.py enrich --apply ``` `scripts/skill-generator/` also ships `README.md` and `SKILL-FORMAT.md`. --- ## Contributing 1. **Add a skill** — use the generator, or create `skills/<name>/SKILL.md` by hand following the format above. 2. **Validate** — run `python -m ci --fast` (or `python -m ci --only skills`). 3. **Commit** — the pre-commit hook runs preflight + static automatically. 4. **Push** — the pre-push hook runs the fast pipeline; it must be green. Bulk additions and upstream imports are welcome as pull requests. --- ## License & Attribution - **MIT** — 10,088 skills, including all adapted skills from MIT-licensed upstreams. - **UPL-1.0** — 19 skills derived from [oracle/skills](https://github.com/oracle/skills) (Copyright (c) 2025 Oracle and/or its affiliates). Adapted skills retain their upstream license and credit the source repository, path, and copyright holder in a `Source & Attribution` section; the CI pipeline requires that attribution and flags missing or altered licenses. --- ## Documentation Map | Document | Contents | | --- | --- | | [docs/skills-index.md](docs/skills-index.md) | Every skill, grouped by category, with descriptions | | [ci/README.md](ci/README.md) | Pipeline stages, CLI, automation, configuration | | [docs/IMPORTING.md](docs/IMPORTING.md) | Upstream import and customization workflow | | [scripts/skill-generator/SKILL-FORMAT.md](scripts/skill-generator/SKILL-FORMAT.md) | Generator-specific format notes | | [Hermes Agent docs](https://hermes-agent.nousresearch.com/docs) | Host agent documentation |

正在加载仓库 README…

这个仓库中的 skills

职业分类待补全

未分类

Use when applying 3pl management implementation.

原文语言：英语

2026-09-24

已展示 40 / 10,021 个已收集 Skill。