# ๐ธ skill-bartender ### *Mix the right skill cocktail for every task โ and never pour an untasted bottle.* [](LICENSE) [](https://github.com/akqwpeter-prog/skill-bartender/actions/workflows/scan.yml) [](https://github.com/topics/dsh-plugin) [](docs/skillspector-report.json) [](README.md#-why) [](README.md#-quick-start) [](docs/lang/README_ZH.md)
Your agent already sees a catalog of skill names and descriptions โ but it **over-pours**: loads too many skills, loads the wrong ones, or misses the one workflow skill that composes the task. **skill-bartender** is the meta-skill that fixes the pour: - ๐ช **Laziness ladder** โ zero skills when plain tools suffice; one skill when one matches; workflow over hand-composed atomics; unsure โ don't load. - ๐ท **Routing table** โ a user-editable taskโskill map (`references/policy.md`) that overrides the defaults. - ๐ **Safe cellar** โ a needed skill missing? Quarantine โ SkillSpector scan โ explicit human approval โ install. Never auto-installs. - ๐ง **Learn** โ loaded-but-unused skills get logged and skipped next time. - ๐งช **Taste test** โ audit installed skills and rewrite weak descriptions into "when-to-use" sentences. [Why](#-why) ยท [What you get](#-what-you-get) ยท [Quick start](#-quick-start) ยท [See it in action](#-see-it-in-action) ยท [Usage](#-usage) ยท [Security model](#-security-model-read-this) ยท [FAQ](#-faq) ยท [Examples](#-examples) ยท [Layout](#-layout) ยท [License](#-license) [**English**](README.md) ยท [**็ฎไฝไธญๆ**](docs/lang/README_ZH.md)
## ๐ Usage
Four ways to use it:
| Way | How | When |
|---|---|---|
| **A. Say the name** | In any session, just say "skill-bartender" | One-off or first-time setup |
| **B. Always-on routing** | Paste the routing table into AGENTS.md | Every task routes through the ladder |
| **C. Request a pour** | "Which skill fits this task?" | Choosing among skills |
| **D. Cellar audit** | "Audit my installed skills" | Taste test: weak descriptions get rewritten |
`skill-bartender` must itself be loaded once (user gesture or task match) โ
it never self-triggers, and never pre-loads "just in case".
## ๐ Security model (read this)
- Skills are **instructions**, and instructions can be adversarial (prompt
injection). SkillSpector is a **filter, not a guarantee**.
- `scripts/` in any skill is **code** โ never executed without human review.
- Human approval is mandatory for every install. **No silent installs, ever.**
- This skill scans itself clean: SkillSpector **0 findings** (score 0 / SAFE)
โ [docs/skillspector-report.json](docs/skillspector-report.json).
- Security policy: [SECURITY.md](SECURITY.md).
## โ FAQ
**Does it auto-install missing skills?**
No. Every download goes to a quarantine dir, gets scanned with SkillSpector,
and is copied into the skills root only after explicit human approval. A
passing scan is a filter, not a guarantee โ prompt injection survives static
scans, so scripts are shown to the human and default-deny.
**What if SkillSpector isn't installed?**
`uv tool install git+https://github.com/NVIDIA/skillspector.git`, or run the
manual checklist in `references/policy.md`.
**Does it work with Claude Code and Codex?**
Yes โ the same SKILL.md installs on all three platforms in ~15 seconds.
**How is this different from DshMarket / dsh-find-plugin / dsh-plugin-autoevo?**
They find, search, and auto-install plugins. skill-bartender adds the
**routing policy** (ladder + routing table) and the **quarantine-then-approve**
discipline. Use it *alongside* the ecosystem, not instead of it.
**How is it evaluated?**
The routing policy ships with a gold-task suite: [docs/eval.md](docs/eval.md).
## ๐ Examples
- [docs/EXAMPLES.md](docs/EXAMPLES.md) โ real routing cases, cellar installs, audits.
- [docs/ROUTING-GUIDE.md](docs/ROUTING-GUIDE.md) โ how to write your own taskโskill rules.
- [docs/eval.md](docs/eval.md) โ gold-task suite for the routing policy.
## ๐บ๏ธ Layout
```
skill-bartender/
โโโ skills/
โ โโโ skill-bartender/
โ โโโ SKILL.md # the skill itself (one file, three platforms)
โ โโโ references/policy.md # user-editable routing table
โโโ docs/
โ โโโ screenshots/how-it-works.png
โ โโโ eval.md # gold-task suite
โ โโโ EXAMPLES.md / ROUTING-GUIDE.md
โ โโโ skillspector-report.json # self-scan: 0 findings
โ โโโ social-preview.png # banner (regenerate via scripts/)
โ โโโ lang/README_ZH.md # ็ฎไฝไธญๆ
โโโ scripts/
โ โโโ make-banner.py # composes docs/social-preview.png
โ โโโ make-diagram.py # composes the how-it-works diagram
โ โโโ validate.py # local structure validation
โโโ cordis.patch.yml / index.js / package.json # DSH bundle manifest
โโโ LICENSE (MIT)
```
## ๐ค Join the DSH plugin ecosystem
DeepSeek Harness developer preview is still in its testing phase for Harness
developers; core plugins and base APIs will keep iterating. We look forward
to exploring the upper limits of intelligence together with developers
worldwide, on top of open-source, open, reusable, and composable infrastructure.
- [dsh-plugin topic](https://github.com/topics/dsh-plugin)
- [Quickstart](https://deepseek-harness.github.io/deepseek-harness/guide/quickstart)
- [DeepSeek Harness repo](https://github.com/deepseek-ai/deepseek-harness)
- Companion executor: [dsh-skill-router](https://github.com/akqwpeter-prog/dsh-skill-router)
> This repo is tagged [`dsh-plugin`](https://github.com/topics/dsh-plugin) and
> listed in the [awesome-dsh-plugin](https://github.com/awesome-dsh-plugin/awesome-dsh-plugin)
> curated list. PRs, issues and translations are welcome.
## ๐ License
[MIT](LICENSE). Ponytail (MIT) is referenced, not bundled โ tribute in the SKILL.md.