MCP Server
Connect AI agents to your Digispot crawl data over the Model Context Protocol — query findings, start crawls, and generate fixes without leaving your editor.
The Digispot MCP server exposes your crawl data to any MCP-compatible AI client. Once connected, an agent can query findings, start crawls, and generate fixes directly — without leaving the chat or copying data across tools.
Two server variants
| Desktop Spider | Cloud Platform | |
|---|---|---|
| Transport | stdio (local process) | HTTP (remote) |
| Scope | Per-project on your machine | Cloud account-wide |
| Tools | 60+ | Varies by plan |
| Auth | Project binding in the Spider app | API key from Settings → API Keys |
| Requires | Spider app open, project loaded | Cloud Starter plan or above |
Use the Desktop Spider server when your crawl data lives locally and you want zero latency between the agent and the database. Use the Cloud server when the crawl runs on Digispot's servers or when multiple team members need access.
Desktop Spider: local stdio server
The Spider app generates the MCP configuration for you. Open the Spider, load a project, then go to Settings → MCP Server (or click Connect AI agent in the project toolbar). Copy the generated config block.
Add it to your client's MCP configuration file:
{
"mcpServers": {
"digispot-spider": {
"command": "/Applications/Digispot Spider.app/Contents/MacOS/mcp-server",
"args": ["--project", "/path/to/your/project.digispot"]
}
}
}Copy the exact path from the Spider app rather than typing it — the binary location varies by macOS/Windows and install path. The Spider app generates the correct block under Settings → MCP Server.
The server exposes 60+ tools covering crawl management, issue queries, page reports, and AI analysis tasks. Representative tools:
| Tool | What it does |
|---|---|
get_site_summary | Returns the current crawl's top-level scores and issue counts |
list_issues | Lists all issues, filterable by category, severity, or URL |
get_page_report | Returns the full audit for a specific URL |
start_crawl | Triggers a new crawl of the loaded project |
get_crawl_status | Returns the current crawl's progress and page count |
run_ai_task | Runs one of the 24 AI analysis tasks against the current crawl |
get_top_pages | Returns the highest-traffic pages ranked by GSC data |
search_pages | Full-text search across page titles, URLs, and issues |
Requires: Spider app running, project open, Tracker or Leader plan for AI tools.
Cloud Platform: HTTP server
Generate an API key under Settings → API Keys in the Cloud Platform, then add the server to your MCP client:
{
"mcpServers": {
"digispot-cloud": {
"command": "npx",
"args": ["-y", "@digispot/mcp-server"],
"env": {
"DIGISPOT_API_KEY": "dsp_live_..."
}
}
}
}Your API key grants full read and write access to your cloud account. Store it in an environment variable or secrets manager — never in version control.
What to ask an agent
Once connected, the agent has access to your real crawl data. Example prompts:
- "What are the top 10 issues on example.com by severity?"
- "Which pages are failing Core Web Vitals on mobile?"
- "Show me all pages missing a meta description and draft fixes for each."
- "Start a new crawl of the project and tell me when it's done."
- "Which pages have the most organic traffic but are also failing structured data checks?"
The agent can combine tool results — for example, pulling the top traffic pages, cross-referencing with failing checks, and generating a prioritized fix plan in one response.