Skip to content
Back to skills

Setup

ASecurity

Bootstrap or repair the skill-concierge engine from scratch. Use this skill when installing skill-concierge on a new machine, right after a plugin update, or when skill-concierge:doctor reports the engine venv is missing. Runs setup.sh to build the stable engine venv, start the local index owner, build the multilingual index, and apply the curated skill-budget overrides, then verifies the result with doctor.

  • 2 stars
  • 0 votes
  • 0 copies
  • 2 views
  • Added October 19, 2026
ai-agentspythonbashsqldockergit

Works with

  • claude code
  • mcp

Security analysis

A100/100

Scanned October 19, 2026

npx -y skills add thinhkhuat/skill-concierge --skill setup --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Setup?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Setup
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/thinhkhuat-setup/badge)](https://www.skillsdirectory.com/skills/thinhkhuat-setup)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: setup
user-invocable: true
description: Bootstrap or repair the skill-concierge engine from scratch. Use this skill when installing skill-concierge on a new machine, right after a plugin update, or when skill-concierge:doctor reports the engine venv is missing. Runs setup.sh to build the stable engine venv, start the local index owner, build the multilingual index, and apply the curated skill-budget overrides, then verifies the result with doctor.
license: MIT
metadata:
  version: 0.1.1
next-skills: skill-concierge:doctor
---

# skill-concierge setup

First-time bootstrap (and post-update refresh) for the vendored skill-search engine.
`setup.sh` is idempotent — safe to re-run any time.

## Prerequisites

- **Python 3.10–3.12** on `PATH` (or set `SKILL_PYTHON=/path/to/python3.12`).
No Docker: the vector store is the local index owner (`python -m skill_search.index_owner`
from the venv, one SQLite file), which `setup.sh` starts itself.

If Python is missing, tell the user and stop — `setup.sh` cannot proceed without it.

## Steps

1. **Run the bootstrap** from the plugin root:

   ```bash
   bash "$CLAUDE_PLUGIN_ROOT/setup.sh"
   ```

   (When working from a git clone instead, `cd` into the repo and run `./setup.sh`.)
   It performs four idempotent steps: stable venv + deps → (re)start the index owner → build/refresh
   the index (discovers skills from every supported harness's skill roots) and the
   actionability-gate corpus → apply curated overrides to `~/.claude/settings.json`
   (backed up first; Claude Code only — a harmless no-op when run from Codex).
   Setup never builds the keep-off map: it is consent-only ([ADR-0077](../../docs/adr/0077-keep-off-map-is-consent-only.md)).

2. **De-duplicate** any prior user-scope MCP so only the bundled one runs:

   ```bash
   claude mcp remove skill-search -s user
   ```

3. **Restart Claude Code** so the MCP server and the settings overrides take effect.

4. **Verify** the install:

   ```bash
   python3 "$CLAUDE_PLUGIN_ROOT/scripts/doctor.py"
   ```

   Expect `status: OK`. In Claude Code, `/mcp` should list `skill-concierge:skill-search`
   as connected. If anything is degraded, hand off to the `skill-concierge:doctor` skill.

## When to re-run

- After **any plugin update** (`/plugin marketplace update` + reinstall) — refreshes the
  engine copy in the stable venv and rebuilds the index.
- When the owner's index file (`~/.cache/skill-search/index.sqlite`) or the venv was wiped.

## Notes

- The venv lives at `~/.claude/skill-concierge/venv` (outside the plugin cache, so it
  survives reinstalls — ADR-0004). Override with `SKILL_CONCIERGE_VENV`.
- `setup.sh` reads the embedder + store URL from `.mcp.json` so the built index can never
  diverge from the model the live MCP uses.

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.