obsidian-find — Claude Code command: smart vault search
obsidian-find searches an Obsidian vault and delivers relevant excerpts with context instead of bare filenames.
MIT ★ 5,412 updated 2026-10-24
/plugin marketplace add eugeniughelbur/obsidian-second-brain
/plugin install obsidian-second-brain@obsidian-second-brainWhat obsidian-find does
The command runs ranked keyword searches across notes with stopword filtering and length normalization. It checks for _CLAUDE.md in the vault root before proceeding. To perform searches, it relies on the obsidian_search MCP tool or vault_ops.search located in integrations/obsidian-mcp-server/. When no specialized search tool is configured in Claude Code, it falls back to grepping files directly, ensuring short entries in wiki/ rank higher than transcripts in raw/ or long log.md files.
Every search hit returns the title, folder, note category, and a matching snippet. If queries return mixed results, matches are grouped by entity type like tasks, people, or projects. Notes written or edited by this workflow must strictly conform to references/ai-first-rules.md, including an ai-first: true frontmatter flag, typed metadata, and compulsory [[wikilinks]]. If that rules file cannot be read from the directory tree, the agent must report it before writing.
What it helps with
- locate vault notes with matching excerpts and folder context
- cluster search hits into people, projects, and task buckets
- prevent long transcripts and logs from overshadowing canonical notes
- verify note absence exhaustively to prevent hallucinated misses
Example prompts for the agent
- /obsidian-find deployment pipeline architecture
- /obsidian-find client feedback on onboarding
- Search my notes for database migration decisions
How to install obsidian-find
1Plugin marketplace
/plugin marketplace add eugeniughelbur/obsidian-second-brain/plugin install obsidian-second-brain@obsidian-second-brain2Manually: put the files in place
.claude/commands/obsidian-find.mdCall it with /obsidian-find. For all projects use ~/.claude/commands/
Manually: put the files in place
~/.codex/prompts/obsidian-find.mdCall it with /prompts:obsidian-find. Codex marks this format deprecated in favour of skills
Manually: put the files in place
.github/prompts/obsidian-find.prompt.mdManually: put the files in place
.gemini/commands/obsidian-find.tomlGemini stores commands as TOML: the prompt text goes into the prompt field
Questions about obsidian-find
How to install obsidian-find in Claude Code?
Add the marketplace source with /plugin marketplace add eugeniughelbur/obsidian-second-brain, then run /plugin install obsidian-second-brain@obsidian-second-brain.
What does obsidian-find do?
It queries your Obsidian vault using length-normalized ranking, returning note categories, paths, and relevant snippets rather than just names.
What happens if no Obsidian MCP server is active?
Claude Code falls back to searching via grep, manually prioritizing canonical notes over noisy logs and raw transcripts.
Source text
obsidian-find.md · MIT
--- description: Smart vault search - returns results with context, not just filenames category: vault triggers_en: ["find in vault", "search my notes", "where is", "what did I write about"] triggers_de: ["im Vault suchen", "meine Notizen durchsuchen", "wo ist", "was habe ich über ... geschrieben"] triggers_es: ["busca en el vault", "busca en mis notas", "dónde está", "qué escribí sobre", "tengo algo sobre"] triggers_pt: ["encontre no vault", "pesquise minhas notas", "onde está", "o que eu escrevi sobre"] triggers_zh: ["在知识库里找一下", "搜索我的笔记", "我之前在哪篇笔记里写过", "我写过关于这个吗"] --- Use the obsidian-second-brain skill. Execute `/obsidian-find $ARGUMENTS`: The argument is the search query. 1. Read `_CLAUDE.md` first if it exists in the vault root 2. Search the vault for the query using the ranked keyword search where it is available: the `obsidian_search` MCP tool, or `vault_ops.search` directly (`integrations/obsidian-mcp-server/`). It applies stopword filtering and length-normalized ranking, so a short note with the term in its title outranks a long note that merely repeats it. In Claude Code (where no search tool is bound), grep the vault and read the top matches directly, applying the same judgement: ignore filler words, and do not let long `raw/` transcripts or `log.md` outrank a canonical `wiki/` note. 3. Also try variations if results are sparse (synonyms, related terms) 4. Return results with context: note title, folder, a relevant excerpt, and what type of note it is 5. If results are ambiguous, group them by type (people, projects, tasks, etc.) 6. Offer to open, update, or link any of the found notes Do not just return filenames - return enough context for the user to act on the results. --- **AI-first rule:** Every note created or updated by this command MUST follow `references/ai-first-rules.md` - `## For future agent` preamble, rich frontmatter (`type`, `date`, `tags`, `ai-first: true`, plus type-specific fields), recency markers per external claim, mandatory `[[wikilinks]]` for every person/project/concept referenced, sources preserved verbatim with URLs inline, and confidence levels where applicable. If that path does not resolve from your working directory, search upward for it; if you still cannot read it, say so before writing rather than producing a note that silently skips the rule. The vault is for future agent retrieval - not human reading. **Anti-fabrication:** Search exhaustively before claiming any note, person, or file is absent - false absence is the most common failure mode - and never invent facts, entities, or dates (mark unknowns as `TBD`). See the anti-fabrication and search-completeness hard rules in `references/ai-first-rules.md`.
© Eugeniu Ghelbur · MIT · eugeniughelbur/obsidian-second-brain