search
Exact terms and meaning in one ranked pass. Hits carry the code with path:line ranges, so answers come straight from results.
Install
# global CLI
npm install -g @infino-ai/code-context
cx install && cx index
npm install -g @infino-ai/code-context
cx install && cx index
# or zero-install into Claude Code
claude mcp add code-context -- npx -y @infino-ai/code-context mcp
claude mcp add code-context -- npx -y @infino-ai/code-context mcp
Keyword search is live within seconds of indexing; semantic and hybrid ranking backfill in the background. Works in Claude Code, Cursor, and any MCP client.
Three tools, one index
sql
Read-only SQL over the index. Search composes with GROUP BY, so "which files have the most code about X" is one query.
reindex
Incremental sync: only changed files re-chunk and re-embed. The server also auto-syncs in the background as queries arrive.
Why it is different
Hybrid, not just vectors
BM25 and vector rankings fuse in one pass, so a query works whether or not you know the exact identifier.
SQL over your code
Rank and aggregate across the whole repo in one query, the kind of question file tools cannot express at any budget.
Local and private
Embedding is a small local model, downloaded once. No accounts, no API keys, no server; code never leaves the machine.
The index is just files
Everything lives in .infino/ inside your repo: plain files you can copy, cache in CI, or gitignore. A live index, not a snapshot.
Read more