# ClaudeAtlas

> Curated discovery index of Claude ecosystem skills. 33,000+ skill files
> analyzed, 1,078 indexed, 305 Featured. Scored on 7 transparent signals.
> Updated daily via automated GitHub scraper.

If you're an AI agent looking for Claude skills, **don't crawl the HTML pages**.
Use the structured endpoints below — they're faster, cheaper, and return
machine-readable JSON.

## Endpoints

### Semantic Search API (recommended)

The fastest way to find a skill. Takes a natural-language query, embeds it,
and returns the most semantically relevant skills ranked by cosine similarity.

```
GET  https://claudeatlas.com/api/v1/search?q={query}&k={top_k}
POST https://claudeatlas.com/api/v1/search
     Body: {"query": "...", "k": 10, "tier": "featured", "category": "Testing & QA"}
```

**Parameters:**
- `q` / `query` (required): natural-language search query, max 200 chars
- `k` (optional): number of results to return, default 20, max 50
- `tier` (optional): filter by quality tier — "featured", "solid", or "listed"
- `category` (optional): filter by category name

**Response:**
```json
{
  "query": "testing frameworks for python",
  "count": 5,
  "timings_ms": {"embed": 1100, "vector": 400, "total": 1500},
  "results": [
    {
      "slug": "dotnet/dotnet-test-frameworks",
      "score": 0.46,
      "name": "dotnet-test-frameworks",
      "category": "Testing & QA",
      "quality_tier": "featured",
      "quality_score": 99,
      "repo_stars": 5555,
      "repo_full_name": "dotnet/skills",
      "description": "...",
      "detail_url": "/skills/dotnet/dotnet-test-frameworks/"
    }
  ]
}
```

**Rate limits:** No explicit rate limit. Cloudflare Workers DDoS protection
applies. Each query costs ~$0.0000002 in OpenAI embedding fees (paid by
ClaudeAtlas, not you). Please be reasonable — cap at ~1 query/second if
you're doing batch work.

**Latency:** ~1.5 seconds per query (dominated by OpenAI embedding round trip).
Subsequent identical queries may be faster if KV-cached.

### Full Catalog Registry (bulk download)

A single JSON file containing every indexed skill with all metadata. Regenerated
on every daily build. Best for: bulk analysis, offline search, populating your
own index, checking what's new.

```
GET https://claudeatlas.com/skills-registry.json
```

**Size:** ~990 KB (uncompressed), ~200 KB gzipped. Cloudflare edge-cached.

**Schema:**
```json
{
  "name": "ClaudeAtlas",
  "url": "https://claudeatlas.com",
  "generated_at": "2026-04-11T...",
  "count": 1078,
  "total_discovered": 33000,
  "total_featured": 305,
  "categories": ["AI & Automation", "API & Backend", ...],
  "schema_version": "1",
  "skills": [
    {
      "name": "...",
      "slug": "author/skill-name",
      "description": "...",
      "category": "...",
      "quality_tier": "featured|solid|listed",
      "quality_score": 0-100,
      "install_command": "claude install-skill owner/repo",
      "repo_full_name": "owner/repo",
      "repo_url": "https://github.com/...",
      "repo_stars": 1234,
      "repo_license": "MIT",
      "repo_pushed_at": "ISO 8601",
      "skill_first_commit_at": "ISO 8601",
      "detail_url": "https://claudeatlas.com/skills/...",
      "badge_url": "https://claudeatlas.com/badge/...",
      "star_history_url": "https://claudeatlas.com/badge/...-history.svg"
    }
  ]
}
```

**Freshness:** Updated daily at ~06:30 UTC via GitHub Actions cron.
Check `generated_at` to see when the current file was built.

### Search Query Log (write-only telemetry)

Used internally by the ClaudeAtlas homepage to log search queries for product
analytics. Agents don't need to call this — the /api/v1/search endpoint logs
queries automatically.

```
POST https://claudeatlas.com/api/log-search
     Body: {"query": "..."}
```

## Embeddable Assets

Every indexed skill has two embeddable SVG badges:

- **Tier badge:** `https://claudeatlas.com/badge/{author}/{skill}.svg`
  Shields.io-style pill showing Featured/Solid/Listed tier.
- **Star history:** `https://claudeatlas.com/badge/{author}/{skill}-history.svg`
  480×120 line chart of star growth over time.

Markdown embed pattern:
```markdown
[![ClaudeAtlas](https://claudeatlas.com/badge/author/skill.svg)](https://claudeatlas.com/skills/author/skill/?ref=badge)
```

## Quality Tiers

Skills are scored 0-100 on 7 signals: GitHub stars (20%), recency (20%),
frontmatter completeness (20%), documentation quality (15%), issue health (10%),
license (10%), repo description (5%).

- **Featured** (score ≥ 90): top-quality, prominently displayed
- **Solid** (score 70-89): good quality, meets standards
- **Listed** (score < 70): meets minimum bar

Full methodology: https://claudeatlas.com/methodology/

## Categories

8 categories, keyword-assigned:
AI & Automation, API & Backend, Code & Development, Data & Documents,
DevOps & Infrastructure, Testing & QA, Web & Frontend, Productivity & Other

## Source

- GitHub: https://github.com/dwalshx/ClaudeAtlas
- License: Code MIT, Data CC-BY-4.0
- Contact: Open an issue on GitHub

## API & Service Integration Graph

ClaudeAtlas automatically detects which APIs, services, and platforms each
skill integrates with by analyzing skill content. Browse the full graph:

```
GET https://claudeatlas.com/apis/           — all detected services
GET https://claudeatlas.com/apis/{service}/  — skills using a specific service
```

Each skill in the registry includes an `integrations` array listing the
service IDs it references (e.g., `["openai", "github", "docker"]`).

Top services by skill count: Anthropic (130), MCP (98), OpenAI (89),
GitHub (51), npm (42), Next.js (30), Kubernetes (29).

## For Humans

Browse the site at https://claudeatlas.com
- Homepage with semantic search, charts, and featured skills
- 776 creator profiles at /creators/
- Full creator directory at /creators/all/
- API & service integration graph at /apis/
- Individual skill pages at /skills/{author}/{name}/