Spiderlook MCP server docs
Connect a Spiderlook project to Claude Desktop, the claude.ai connector, or ChatGPT. The assistant reads rivals, activity, and evidence from the MCP server below.
MCP endpoint
Use this URL in every client. Authentication is OAuth. The client opens Spiderlook so you can sign in and approve. Do not paste a bearer token or an API key.
https://api.spiderlook.com/mcp
The server name is spiderlook. The path accepts a trailing slash. Access tokens are for this resource only.
Setup
Claude Desktop
- Add a remote MCP server named spiderlook.
- Set the URL to https://api.spiderlook.com/mcp.
- Or run claude mcp add --transport http spiderlook https://api.spiderlook.com/mcp.
- When Claude asks you to sign in, approve access in Spiderlook. Do not paste a bearer token.
claude.ai connector
- Open Claude on the web and go to Customize, then Connectors.
- Add a custom connector named Spiderlook and paste the MCP endpoint.
- The prefilled connector link opens that modal with the name and URL filled in.
- Choose Connect. Claude starts OAuth and opens Spiderlook so you can sign in and approve.
ChatGPT
- In ChatGPT, add a custom connector or app that accepts a remote MCP server URL.
- Name it Spiderlook and set the server URL to https://api.spiderlook.com/mcp.
- Create the connector. ChatGPT starts OAuth and opens Spiderlook so you can sign in and approve.
- Do not paste a bearer token. Spiderlook does not deep-link ChatGPT, because ChatGPT does not provide a connector install link.
Tools
These are the tools the Spiderlook MCP server registers. Each line is the tool’s own description. Call list_workspace before the others, and copy project_id and competitor_id from that result.
list_workspace- Catalog of orgs, projects, and competitors the user can access.
query_metrics- Batched project counts and named formulas. Not a dashboard narrative.
query_activity- Scraped posts, ads, or pages for one channel, windowed and newest first.
get_item- Full stored row, current profile, or snapshot series.
list_events- Persisted deviations or strategy versions, newest first, cursor-paged.
add_competitor- Create a competitor from a website URL. May cost extra.
update_competitor- Re-label kind (competitor ↔ trendsetter) or toggle continuous_tracking.
add_channel- Attach a channel on an existing competitor and queue ingest. May cost extra.
get_battlecard- The team's sales battlecard for one rival.
generate_battlecard- Regenerate a rival's battlecard from its analysed data. May cost extra.
review_battlecard_suggestion- Accept or dismiss one pending battlecard suggestion.
approve_battlecard- Mark a rival's battlecard as approved by the caller.
Read tools return scraped posts, ads, pages, profiles, counts, and the sales battlecard. Write tools that add a rival, attach a source, or regenerate a battlecard may cost extra.
Example prompts
- List the projects and rivals I can access.
- For this project, who had the most activity in the last 30 days? Use activity_counts and relative_activity_index.
- Show the newest Meta ads for this rival.
- Open the full stored row for this post, including comments where they are stored.
- Which playbook-break deviations are still open for this rival?
- Show this rival’s sales battlecard: summary, why we win, why they win, and open objections.
FAQ
The server is https://api.spiderlook.com/mcp. It speaks streamable HTTP, and a trailing slash on that path is accepted.
The registered server name is spiderlook. Clients sign in with OAuth. Do not paste a bearer token.
Guides
How competitor research fits in ChatGPT and Claude, and how Spiderlook compares with other MCP servers people search for.