tomevault-io mgrep — semantic search
mgrep searches by meaning, not exact text. Use it for intent and concept search; use grep/ripgrep for exact matches and symbol tracing.
When to use
| Use mgrep for | Use grep/ripgrep for |
|---|---|
| "Where is auth configured?" | Exact symbol or identifier |
| "How do we parse chunks?" | Refactoring, rename |
| Feature discovery, onboarding | Regex, known patterns |
How to run
From the workspace root:
mgrep -c -m 15 "<query>" [path]
-c/--content: include snippet content (recommended for agent use).-m <n>: max results (e.g. 15–25).- Optional
path: e.g.src/orsrc/gitUtils.
Examples:
mgrep -c -m 15 "where do we set up auth?"
mgrep -c -m 15 "branch checking logic" src/gitUtils
Optional flags
--answer/-a: summarized answer from results.--web/-w: include web search (docs, tutorials).--agentic: break complex questions into sub-queries.--sync/-s: sync store before searching (if index may be stale).
Using results
Rely on mgrep output (paths and snippets) to answer or edit; do not invent file contents. If needed, read the reported files for full context.
Setup (if mgrep unavailable)
- Install:
npm install -g @mixedbread/mgrep(or pnpm/bun). - Auth:
mgrep loginor setMXBAI_API_KEY. - Index (optional):
mgrep watchin project root.
Converted and distributed by TomeVault — claim your Tome and manage your conversions.