infino-ai/supergrep

code-context

When your agent uses it

How to answer codebase questions with the SuperGrep MCP tools: find and plain sql locally (exact-text lookup that replaces grep, read-only SQL aggregation over the index), and search and ask over the platform copy when the server has an account (ranked hybrid keyword+semantic search, and small models that retrieve the facts instead of you crawling the repo yourself). Use when you would grep for an identifier or literal, when a question spans many files ("how does X work", "where is Y handled"), when ranking or counting code by topic across a repo, or when the code-context tools are present but deferred and need loading before use.

What it loads into context

Token counts are estimates. Scripts run separately; they are not counted as reference context.

# code-context: search over the repository

code-context maintains a local index of the repository (in `.infino/` at the
repo root) and always exposes `find`, `sql` and `read` over it; `search` and
`ask` register too when the server has an account, running over the same
index's platform copy where the embeddings are computed. Every lookup an
agent would otherwise do with grep or by crawling files runs against the
index instead, or is delegated to small models entirely: `find` for the
exact-text case, `search` for one ranked pass over everything that spans
the repo, `ask` when you would rather hand the retrieval off than do it
yourself.

## If the tools are deferred