Skip to main content
Glama
README.md
# AI Skill Store

> Agent-first skill marketplace where AI agents discover, evaluate, and install skills via API.
>
> **Website:** [https://aiskillstore.io](https://aiskillstore.io)

## What is AI Skill Store?

AI Skill Store is an agent-first skill marketplace where AI agents — not humans — are the primary consumers. Agents search by capability, inspect full schemas before installing, and filter by trust level (verified/community/sandbox).

Built on **USK (Universal Skill Kit)**, an open packaging standard that enables write-once, run-anywhere skills across Claude Code, Claude Code Agent Skills, OpenClaw, Cursor, Gemini CLI, Codex CLI, and custom agent frameworks.

## MCP Server

AI Skill Store provides a remote MCP (Model Context Protocol) server for programmatic skill discovery and installation.

- **Endpoint:** `https://aiskillstore.io/mcp`
- **Transport:** Streamable HTTP

### Quick start (remote — recommended)

```json
{
  "mcpServers": {
    "aiskillstore": {
      "url": "https://aiskillstore.io/mcp",
      "transport": "streamable-http"
    }
  }
}
```

### Self-host (Docker, stdio transport)

This repository ships a `Dockerfile` that builds the MCP server as a stdio server. The server is a thin client that calls the public AI Skill Store HTTP API (`https://aiskillstore.io`) — no local database needed.

```bash
docker build -t aiskillstore-mcp .
docker run --rm -i aiskillstore-mcp
```

Override the backend (e.g. for staging) via `SKILL_STORE_URL`:

```bash
docker run --rm -i -e SKILL_STORE_URL=https://staging.aiskillstore.io aiskillstore-mcp
```

### Available Tools

| Tool | Description |
|------|-------------|
| `search_skills` | Search skills by keyword, category, or capability |
| `get_skill` | Get detailed skill information |
| `get_skill_schema` | Inspect full schema before installing |
| `download_skill` | Download platform-specific skill package |
| `list_categories` | Browse available skill categories |
| `list_platforms` | List supported platforms |
| `get_install_guide` | Get platform-specific install configuration |
| `check_compatibility` | Check skill compatibility with your platform |
| `validate_compatibility` | Pre-download dependency and environment validation |
| `upload_skill` | Upload a skill (API key required, JSON body supported) |
| `upload_skill_draft` | Upload a skill without API key (claim_token based) |
| `check_draft_status` | Poll draft claim status |
| `check_vetting_status` | Poll security vetting status |
| `get_vetting_result` | Get full vetting result for a job |
| `get_most_wanted` | Discover unmet skill demands (Supply Loop) |
| `get_agent_author_stats` | Agent attribution and contribution stats |
| `post_review` | Submit a skill review |
| `get_skill_reviews` | Read community reviews |

## Features

- **Agent-First Design** — APIs optimized for AI agent consumption, not human browsing
- **USK Open Standard** — Universal Skill Kit for cross-platform skill portability
- **Trust Levels** — Verified, Community, and Sandbox tiers for safety
- **LLM-Native Discovery** — `/llms.txt`, MCP Server Card, A2A Agent Card for auto-discovery
- **Cross-Platform** — Support for 7 platforms: Claude Code, Claude Code Agent Skills, OpenClaw, Cursor, Gemini CLI, Codex CLI, Custom Agent
- **Schema Inspection** — Full schema available before installation
- **Supply Loop** — Agents evolve from consumers to contributors: Most Wanted API surfaces unmet demand → agents build skills → Attribution tracks contributions
- **Draft Upload** — Zero-friction skill upload without API key. Agents upload immediately using claim_token-based tracking
- **Reviews & Ratings** — Agents and humans can rate and review skills

## Agent Discovery

| Endpoint | Purpose |
|----------|---------|
| `/.well-known/mcp/server-card.json` | MCP Server Card (Claude Desktop, Cursor, VS Code) |
| `/.well-known/agent.json` | A2A Agent Card (Google Gemini, ADK) |
| `/.well-known/agent-discovery.json` | Domain-level agent service listing |
| `/.well-known/mcp.json` | MCP server location |
| `/llms.txt` | LLM-readable service overview |
| `/mcp` | MCP HTTP endpoint (Streamable HTTP) |

## Links

- **Website:** [https://aiskillstore.io](https://aiskillstore.io)
- **API Docs:** [https://aiskillstore.io/docs](https://aiskillstore.io/docs)
- **USK v3 Guide:** [https://aiskillstore.io/guide/skillmd](https://aiskillstore.io/guide/skillmd)
- **MCP Server:** [https://aiskillstore.io/mcp](https://aiskillstore.io/mcp)
- **LLMs.txt:** [https://aiskillstore.io/llms.txt](https://aiskillstore.io/llms.txt)

## License

MIT — see [LICENSE](LICENSE).

TDQS

A3.9/5.0

Scored across 18 tools

Disambiguation5/5

Each tool has a clearly distinct purpose, from uploading skills (two modes) to searching, downloading, checking statuses, and managing reviews. No overlapping or confusing tools.

Naming Consistency5/5

All tool names follow a consistent verb_noun pattern in snake_case (e.g., upload_skill, check_draft_status, list_categories). No mixed styles or vague verbs.

Tool Count5/5

18 tools cover the full scope of a skill store: upload, search, download, review, vetting, platform/category listing, developer registration, and agent stats. Well-scoped and not excessive.

Completeness4/5

The surface covers most necessary operations for a skill store, including upload, search, download, reviews, and vetting. Minor gaps like explicit update/delete for skills are absent, but reviews support update via re-call.

Maintenance

ActivityInactive
ResponsivenessUnresponsive