vault-retrieve
Skill0 starsChanged 3 months agoMITView on GitHub
Answer a question from the vault — generate the query fan-out, read every match, and return the relevant content verbatim with its paths. Use when Shane says "what do my notes say about X", "search the vault for X", "pull everything on X", "check my notes before we start", or when any task needs vault context before it can proceed. Do NOT use to write or update notes (use /obsidian or /ingest), to find links between two known concepts (use /connect), or to trace how an idea changed over time (use /trace).
In this repository17 other files it gives its agents.
- backlinksSkill
- challengeSkill
- connectSkill
- expandSkill
- fix-nested-code-fencesSkill
- ghostSkill
- ingestSkill
- learnedSkill
- mapSkill
- meetingSkill
- obsidianSkill
- source-mineSkill
- traceSkill
- vault-indexSkill
- vault-lintSkill
- weekly-learningsSkill
- weekly-signalsSkill
The file
We split this skill into the rules it gives an agent
2 rules. About 0 are common, 0 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 43
Default to Knowledge/. Include Archive/ only when asked about history — it holds superseded material that reads as current.
- Very specific: line 45
Never reach vault files through Read/Grep/Glob or a direct bash path.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Skill: /vault-retrieve [topic]
Return what the vault actually says about a topic — pick the queries yourself, read the hits, and quote them, so the asker never has to hand over a list of search commands.
Don't: run one broad query and stop; the matcher is fuzzy and returns a near-identical set for any broad phrasing, so one query means one blind spot. Don't report a filename as an answer — a hit is a candidate, and the note has to be read before you can say what it holds. Don't paraphrase when the asker needs the source; return the text.
Steps
- Read vault-map.md for folder layout and query heuristics. It is why this skill exists — hand-written retrieval prompts kept re-supplying the map.
- Generate 3–5 narrow queries, differently worded, covering:
- the topic's own vocabulary
- its aliases and former names (projects get renamed — LeadSurface material is also filed under
signalbloomandLede) - an adjacent framing likely used in a title
obsidian search query='<term>' limit=10 - Read every plausible hit —
obsidian read path='<exact path>'(orfile='<name>'without.md). Read in full. A range read is how a note gets misreported. - Return the content, grouped by note, each with its path. Verbatim for anything the asker will quote, act on, or check. Summarize only the clearly peripheral.
- State the gaps. Name what you searched that returned nothing. "Nothing in the vault covers X" is a real answer and more useful than a stretched match.
Output
## <topic> — N notes
### <path/to/Note.md>
<relevant content, verbatim>
### <path/to/Other.md>
<relevant content, verbatim>
## Not found
Searched: <queries that returned nothing useful>Scope guards
- Default to
Knowledge/. IncludeArchive/only when asked about history — it holds superseded material that reads as current. Daily/is episodic. Search it for "when did I do X", not for what Shane thinks about X.- Never reach vault files through
Read/Grep/Globor a direct bash path. TheobsidianCLI is the only supported access path.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

