Live MCP status
probed 2026-09-05 15:39 · 18/21 healthy ·
262 tools
Measured by a real stdio handshake against each server, not read from the config.
A server listed in the catalogue below can still be down — this table is the
authority.
| Server | State | Tools |
If down, why |
|---|
| agent-search | live | 4 | — |
| brightdata-mcp | live | 5 | — |
| browser-tools-mcp | live | 14 | — |
| browser-use | live | 16 | — |
| docker-services-mcp | live | 1 | — |
| exa-mcp-server | live | 2 | — |
| fairchoice-seo-mcp | live | 3 | — |
| firecrawl-mcp | live | 26 | — |
| github-mcp-server | live | 30 | — |
| jina-reader | live | 1 | — |
| mdla-hands | live | 17 | — |
| mdla-marketing | live | 8 | — |
| playwright-mcp | live | 24 | — |
| scrapling | live | 3 | — |
| search-console-mcp | live | 98 | — |
| tavily-mcp | live | 5 | — |
| teachable-scraper | live | 1 | — |
| wet | live | 4 | — |
| agentql-mcp | down | 0 | Needs AGENTQL_API_KEY — not supplied yet. |
| gsc-mcp | down | 0 | Package bug — rejects its own tool schema (Zod validation). Search Console access still works via search-console-mcp (98 tools). |
| serpapi-mcp | down | 0 | Misconfigured: it is an HTTP server that binds 0.0.0.0:8000, but it is registered as a stdio MCP — and :8000 is already taken by skyvern-api. BrightData's search_engine now covers this. |
AI SEO Research Tools
12 picks installed 2026-07-23 · Back to SEO Dashboard · Fable Tools
MCP Servers (configured in OpenClaw)
npx @brightdata/mcp
/usr/lib/node_modules/@brightdata/mcp
npx -y tavily-mcp@latest
/usr/lib/node_modules/tavily-mcp
npx -y exa-mcp-server
/usr/lib/node_modules/exa-mcp-server
/opt/serpapi-mcp/src/server.py
git cloned + pip: serpapi
npx -y agentql-mcp
/usr/lib/node_modules/agentql-mcp
Python Libraries
MarkItDown
microsoft/markitdown ⭐123k
Convert any doc/HTML/PDF to LLM-friendly Markdown — the content ingestion backbone by Microsoft.
pipPython env
markitdown 0.1.6
pip3 install markitdown
scrapegraphai 2.1.5
pip3 install scrapegraphai
Trafilatura
adbar/trafilatura ⭐6k
Battle-tested Python web text + metadata extraction. Top benchmarks for clean content.
pipPython env
trafilatura 2.1.0
pip3 install trafilatura
agentql 1.19.0
pip3 install agentql
CLI / npm Tools
Defuddle
kepano/defuddle ⭐8k
Extract the main content of any page as clean Markdown. From Obsidian creator.
npmCLIsafeBin
defuddle
npm install -g defuddle
@builder.io/gpt-crawler
npm install -g @builder.io/gpt-crawler
Self-Hosted / Cloud API
Jina Reader
jina-ai/reader ⭐11.7k
URL LLM-friendly Markdown (r.jina.ai) + Web Search (s.jina.ai). Self-hostable Docker or free cloud API. OpenClaw MCP configured for hosted endpoint.
cloud APIOpenClaw MCP
https://r.jina.ai/
/opt/jina-reader/ (source) · docker-compose available
API Keys: Click "Get API Key →" on each MCP card above to grab your key. Then set it as an env var and restart OpenClaw.
OpenClaw Integration
MCP Servers Added
brightdata-mcp, tavily-mcp, exa-mcp-server, serpapi-mcp, agentql-mcp, jina-reader
Configured in openclaw.json
/root/.openclaw/openclaw.json -> mcp.servers
Safe Bins + Agent Permissions
defuddle added to safeBins. All MCP tools added to seo-orchestrator agent's alsoAllow.
Configured
/root/.openclaw/openclaw.json
Python Libs in Global Env
markitdown, scrapegraphai, trafilatura, agentql — usable by any OpenClaw Python MCP or agent script.
Available
pip3 list | grep -E "markitdown|scrapegraph|trafilatura|agentql"