<div align="center">

<img src="docs/social-preview.png" alt="skill-bartender — task-to-skill pairing for DeepSeek Harness" width="100%">

<br>

🍸 skill-bartender

Mix the right skill cocktail for every task — and never pour an untasted bottle.

      !Docs

<br>

Your agent already sees a catalog of skill names and descriptions — 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:

• 🪜 Laziness ladder — zero skills when plain tools suffice; one skill

when one matches; workflow over hand-composed atomics; unsure → don't load.

• 🍷 Routing table — a user-editable task→skill map (references/policy.md)

that overrides the defaults.

• 🔐 Safe cellar — a needed skill missing? Quarantine → SkillSpector scan

→ explicit human approval → install. Never auto-installs.

• 🧠 Learn — loaded-but-unused skills get logged and skipped next time.
• 🧪 Taste test — audit installed skills and rewrite weak descriptions

into "when-to-use" sentences.

[Why](#-why) · [What you get](#-what-you-get) · [Quick start](#-quick-start) · [See it in action](#-see-it-in-action) · [Usage](#-usage) · [Security model](#-security-model-read-this) · [FAQ](#-faq) · [Examples](#-examples) · [Layout](#-layout) · [License](#-license)

English · 简体中文

</div>

---

🤔 Why

Most agents treat the skill catalog as an all-you-can-eat buffet. skill-bartender treats it as a bar with a taste test:

Why the "laziness ladder"? A wrong skill body stays in conversation history forever; a missed load only costs one tool round-trip. The best load is the load never made (spirit: ponytail).

✨ What you get

⚡ Quick start

One file, three platforms:

# DeepSeek Harness
mkdir -p ~/.dsh/skills/skill-bartender
cp skills/skill-bartender/SKILL.md ~/.dsh/skills/skill-bartender/
cp -r skills/skill-bartender/references ~/.dsh/skills/skill-bartender/

# Claude Code
mkdir -p ~/.claude/skills/skill-bartender
cp skills/skill-bartender/SKILL.md ~/.claude/skills/skill-bartender/

# Codex
mkdir -p ~/.codex/skills/skill-bartender
cp skills/skill-bartender/SKILL.md ~/.codex/skills/skill-bartender/

Or install as a DeepSeek Harness bundle:

dsh plugin --profile web add github:akqwpeter-prog/skill-bartender

Then say "skill-bartender" once, or paste the routing table into your AGENTS.md for always-on routing. Full examples: [docs/EXAMPLES.md](docs/EXAMPLES.md).

📸 See it in action

The pour flow in one picture: stop at the first rung that holds, and never install without a taste test.

<img src="docs/screenshots/how-it-works.png" alt="How the pour works: laziness ladder (0 plain tools, 1 one match, 2 workflow, 3 unsure) plus the safe cellar (quarantine → SkillSpector scan → human approval → install)" width="100%">

🚀 Usage

Four ways to use it:

skill-bartender must itself be loaded once (user gesture or task match) — it never self-triggers, and never pre-loads "just in case".

🔐 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 — 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)

— [docs/skillspector-report.json](docs/skillspector-report.json).

• Security policy: [SECURITY.md](SECURITY.md).

❓ 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 — 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 — 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).

🎁 Examples

• [docs/EXAMPLES.md](docs/EXAMPLES.md) — real routing cases, cellar installs, audits.
• [docs/ROUTING-GUIDE.md](docs/ROUTING-GUIDE.md) — how to write your own task→skill rules.
• [docs/eval.md](docs/eval.md) — gold-task suite for the routing policy.

🗺️ Layout

skill-bartender/
├── skills/
│   └── skill-bartender/
│       ├── SKILL.md             # the skill itself (one file, three platforms)
│       └── references/policy.md # user-editable routing table
├── docs/
│   ├── screenshots/how-it-works.png
│   ├── eval.md                  # gold-task suite
│   ├── EXAMPLES.md / ROUTING-GUIDE.md
│   ├── skillspector-report.json # self-scan: 0 findings
│   ├── social-preview.png       # banner (regenerate via scripts/)
│   └── lang/README_ZH.md        # 简体中文
├── scripts/
│   ├── make-banner.py           # composes docs/social-preview.png
│   ├── make-diagram.py          # composes the how-it-works diagram
│   └── validate.py              # local structure validation
├── cordis.patch.yml / index.js / package.json   # DSH bundle manifest
└── LICENSE (MIT)

🤝 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
• Quickstart
• DeepSeek Harness repo
• Companion executor: dsh-skill-router

> This repo is tagged dsh-plugin and > listed in the awesome-dsh-plugin > curated list. PRs, issues and translations are welcome.

📄 License

[MIT](LICENSE). Ponytail (MIT) is referenced, not bundled — tribute in the SKILL.md.