← ClaudeAtlas

read-searchlisted

Plain-text (substring) search inside a vault, with surrounding context. EN triggers: "find <text> in my vault", "search for the literal string X", "grep for X", "where do I mention X", "find files containing X". FR triggers : "trouve <texte> dans mon vault", "cherche la chaîne X littéralement", "grep <X>", "où est-ce que je mentionne X", "trouve les fichiers contenant X". Example / Exemple: EN: "find 'risk management' in my vault" FR: "cherche 'gestion du risque' dans mes notes"
tboome33/obsidian-mcp-router · ★ 1 · AI & Automation · score 72
Install: claude install-skill tboome33/obsidian-mcp-router
# read-search Call the obsidian-router `search` MCP tool with arguments parsed from $ARGUMENTS. ## Arguments **Required**: - `query` — the substring to find. **Optional**: - `vault` — vault name. Omit for default. Pass `*` to fan-out across ALL vaults in parallel. - `contextLength` — number of characters of surrounding context per match. Default 100. ## Argument parsing from $ARGUMENTS - the entire $ARGUMENTS as `query` is fine for the most common case - `vault=X query="some words"` or `vault=* query="words"` - if user includes `--all` or "across all vaults", set `vault` to `*` ## Output format Render results as a markdown list, one per matching file: ``` <path> (vault name if cross-vault) - <text snippet with surrounding context> ``` If many matches (>20), summarize counts and show the first 20. Suggest `read-search-smart` (semantic) if the user seems to be looking by meaning rather than literal text.