| name | scout-search |
| description | Use Scout for repo-wide code search in any language - ranked fuzzy discovery, exhaustive strict literal, regex, and word search, plus structural patterns, symbol outlines, and semantic queries. |
Scout Code Search Workflow
Use Scout's find as the primary workspace search instead of shell grep, rg, or find. If Glider (C#) or TGlider (TypeScript and JavaScript) is installed, prefer it for its language, and use Scout for every other language and as the universal fallback.
target: auto routes by query shape, and every response echoes the resolved route in data.route. Use strict match modes - literal, regex, or word - when you need exhaustive grep semantics for an audit or a rename survey; fuzzy is ranked top-N with frecency and is never exhaustive. A zero-hit strict search returns an honest empty result rather than silently falling back, so treat it as evidence. Use target: structural for ast-grep patterns ($NAME for one node, $$$ for many), target: symbols for a tree-sitter declaration outline, and target: semantic for natural-language questions. Outside a shipped grammar, or without a semantic model, these degrade to lexical search and say so in meta.degraded - check server_status.tiers before relying on them.
Narrow with path prefixes, include and exclude globs, and language scoping before widening the query, and page through results instead of loosening the match. Results carry workspace-relative paths, 1-based line and column, and trimmed previews, never whole files. Use server_status for index, watcher, and tier health, and sync {full: true} only to force a full rescan, since ordinary queries already flush pending changes.
Scout never mutates workspace source - its only writes are its own state under .glider/scout/ and ~/.glidermcp. When a hit needs semantic follow-up or an edit, hand the file and span to Glider for C#, or TGlider for TypeScript and JavaScript.