How to give Cursor context of your whole codebase (MCP)

Cursor's index finds code that looks like your question. It doesn't know which function calls which, which tests cover a change, or that a strange retry loop was added after an outage two months ago. Those gaps are where an agent edits the wrong place or undoes a decision made on purpose.

Connect your GitHub repository at asksynapse.in, create a token in Settings → AI agents, then add Synapse to .cursor/mcp.json in your project as an MCP server with your token in the Authorization header. The docs page has the exact JSON. Cursor then lists the Synapse tools in its MCP settings.

• plan_change: before coding, the files and line ranges to change, callers, the pattern to copy and the tests to run.
• check_change: before finishing, broken callers and duplicated code.
• why: the pull requests, reviews and commits behind a piece of code.
• tests_for and blast_radius: which tests cover a file, and what depends on a function.

Larger repositories, shared functions with many callers, and teams with history worth keeping. On a tiny project the agent can read everything anyway. All tools are read-only and Synapse never trains on your code.

In our benchmark (Claude Code with Sonnet, 7 real tasks on a repository Synapse had never seen, 2 runs each), Synapse cut tokens by 58% and cost by 48%, and every edit still compiled. Planning tasks saved the most. Results vary by repository and task.

Synapse is free to start: connect a GitHub repository and add it to Claude Code, Cursor or Claude in about two minutes. Create an account or read the docs.