Skip to content

Claude Code

Register the server with claude mcp add (everything after -- is the server command):

Terminal window
claude mcp add --transport stdio doc-scraper -- /path/to/doc-scraper mcp-server -config /path/to/config.yaml

This uses the default local scope: the server is available to you in the current project only. Add --scope user to make it available in all your projects.

Add --scope project to the command above and Claude Code writes a .mcp.json at the project root that you can commit:

{
"mcpServers": {
"doc-scraper": {
"type": "stdio",
"command": "/path/to/doc-scraper",
"args": ["mcp-server", "-config", "/path/to/config.yaml"],
"env": {}
}
}
}

Claude Code asks each person to approve project-scoped servers from .mcp.json before first use in an interactive session.

Terminal window
claude mcp get doc-scraper

Inside a session, /mcp shows the server’s status and tools.

Ask Claude to orient itself and fetch a corpus, for example:

Call describe_server from doc-scraper. If rust_cli_book has never been crawled, crawl it and wait for the job to finish. Then search it for “error handling” and summarize the best page.

That prompt leads Claude through describe_server, crawl_site, get_job_status, search_docs, and read_page. Once the crawl exists, later sessions can go straight to search.

See the Claude Code MCP docs for scopes and approvals. Checked against the Claude Code docs and claude 2.1.286 on 2026-10-10.