vault-query
Skill3 starsChanged 7 months agoApache-2.0View on GitHub
Karpathy-style vault query and synthesis system. Reads index.md to locate articles, synthesizes answers from multiple sources, and optionally writes high-value results back as new notes. Triggers: "/vault-query", "search vault", "find in notes", "query knowledge base"
3 other projects use a version of this.
Adopted by
Most recently taken up by
- bohanyt/hermes-custom-skillstheir own version, first seen Oct 25, 2026
- Lucioway/way-stacktheir own version, first seen Aug 31, 2026
- quietdaemon/starter-skillstheir own version, first seen Jun 6, 2026
Other versions3 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 3
- bohanyt/hermes-custom-skills★ 1same name
- quietdaemon/starter-skills★ 1same name
- Lucioway/way-stack★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository5 other files it gives its agents.
- article-to-markdownSkill
- chat-to-obsidianSkill
- github-syncSkill
- review-schedulerSkill
- vault-lintSkill
The file
We split this skill into the rules it gives an agent
5 rules. 0 are common, 5 are distinctive, and 0 are very specific to this project.
0
Common
Most of the 3 edited versions we read keep these.
5
Distinctive
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 77
After synthesizing, evaluate whether writing back is warranted:
- Distinctive: line 96
Search for bidirectional links per Bidirectional Link Search
- Distinctive: line 97
Write the new note with complete frontmatter:
- Distinctive: line 113
If written back, report the save location
- Distinctive: line 114
If knowledge gaps were found, suggest topics to add
“Common” is measured: kept by at least 6 in 10 of the 3 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Vault Query — Karpathy-Style Knowledge Synthesis
Based on the "compile, don't retrieve" philosophy: look up the index first, read full text to synthesize, write back high-value answers.
Workflow
Step 1: Environment Detection & Query Parsing
1a. Detect vault root — follow ../../_shared/common-steps.md "Vault Detection" section.
1b. Parse user query intent:
Extract from user input:
query: core question or keywordsscope: whether a directory scope was specified (e.g., user mentions a topic that maps to a configured directory)depth: simple lookup ("which article covers X") vs deep synthesis ("compile everything about X")
Step 2: Index Lookup
2a. Read index.md:
Read("$VAULT_ROOT/index.md")Scan all entries in index.md — titles, summaries, and tags — to find candidate articles related to the query (typically 3-8).
2b. Supplementary search:
index.md may miss content. Use Grep to supplement:
Grep("{keyword}", path="$VAULT_ROOT", glob="**/*.md")Exclude results from .claude/, .git/, reviews/, index.md.
2c. Determine reading list:
Merge index lookup and Grep results, sort by relevance, finalize the list of articles to read in full.
Step 3: Deep Read & Synthesis
3a. Read relevant articles in full:
For each article on the reading list, use Read to get full text. Read in parallel for efficiency.
3b. Synthesize the answer:
Choose synthesis mode based on depth:
- Simple lookup: list relevant articles and their key points, point the user to read directly
- Deep synthesis:
- Extract key information from multiple articles
- Integrate into a complete knowledge picture, citing sources (
see [[article-name]]) - Surface connections, contradictions, or complementary relationships between articles
- Identify knowledge gaps ("no article covers aspect Y of topic X")
Step 4: Write-Back Decision
After synthesizing, evaluate whether writing back is warranted:
Worth writing back:
- Synthesized new insights across 3+ articles
- Discovered undocumented connections between articles
- The user's question reveals a knowledge gap worth filling
- Produced a reusable comparison analysis or decision matrix
Not worth writing back:
- Simple lookup ("which article covers X")
- Answer comes entirely from a single article
- One-off question
4a. If worth writing back, use AskUserQuestion to ask:
"This query produced valuable synthesized knowledge: {brief description}. Save as a new note? Suggested title: {title}, directory: {dir}/"
Upon user confirmation:
- Determine directory per the Directory Mapping Rules in ../../_shared/common-steps.md
- Search for bidirectional links per Bidirectional Link Search
- Write the new note with complete frontmatter:
title: synthesis note titledate: today's datetags: 3-8 relevant tagstype: inferred per Type Inference Rules in ../../_shared/common-steps.mdconfidence: cross-article synthesis is typicallymediumrelated: source article list ("[[article-name]]"format)summary: 1-3 sentence synthesis summarysource: "Vault Query synthesis"
- Update index.md: per the Index Update steps in ../../_shared/common-steps.md, append the new article to the corresponding section
Step 5: Report
Output to user:
- Synthesized answer (with
[[wikilink]]citations) - List of referenced articles
- If written back, report the save location
- If knowledge gaps were found, suggest topics to add
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.

