llmwiki-query
Skill383 starsChanged 5 months agoMITView on GitHub
Answer a question by querying the user's llmwiki. Use when the user asks about their own past work — "what did I decide about X", "what have I been working on", "how did I solve Y", "what's my preferred approach to Z", or any question that the wiki (built from their session history) might answer. Always read the wiki first before falling back to external search.
In this repository7 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- llmwiki-ingestSkill
- llmwiki-syncSkill
- project-maintainerSkill
- self-learnSkill
- wiki-allSkill
The file
We split this skill into the rules it gives an agent
9 rules. About 0 are common, 9 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈9
Distinctive · estimate
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 22
Do NOT invoke for questions that can be answered without the user's personal knowledge base (e.g., "how does TCP work?" → answer from …
- Distinctive: line 26
Follow the Query Workflow from the repo's CLAUDE.md:
- Distinctive: line 44
If yes, write it using the Synthesis Page Format from CLAUDE.md.
- Distinctive: line 53
If the wiki has no relevant pages:
- Distinctive: line 57
If the topic is unlikely to be in session history, offer to answer from general knowledge (and make that explicit: "this is general …
- Distinctive: line 61
Don't make up wiki pages that don't exist.
- Distinctive: line 62
Don't confuse general knowledge with wiki content in the answer.
- Distinctive: line 63
Don't over-cite — if the whole answer comes from one page, cite once at the top.
- Distinctive: line 64
Don't write a wall of text.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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
llmwiki-query
What this skill does
Reads the user's local llmwiki (llmwiki/wiki/) and synthesises an answer to their question, with [[wikilink]] citations pointing to the source pages.
When to use
Invoke this skill when the user asks a question about their own work or history. Examples:
- "What did I decide about the AiShorts backend?"
- "What MCP servers do I have configured?"
- "How did I fix the X bug last week?"
- "What's my preferred way to ingest sessions?"
- "What have I been working on in the germanly project?"
Do NOT invoke for questions that can be answered without the user's personal knowledge base (e.g., "how does TCP work?" → answer from general knowledge).
Workflow
Follow the Query Workflow from the repo's CLAUDE.md:
- Locate the llmwiki install (see
llmwiki-syncskill for the fallback search order). - Read
wiki/index.mdto identify the set of pages that might be relevant. - Read
wiki/overview.mdto get the living synthesis context. - Pick the most relevant pages (sources, entities, concepts) based on the question.
- Read those pages with the Read tool.
- Synthesise an answer with inline
[[wikilink]]citations. Quote directly from source pages when the wording matters. - If the answer is substantial (3+ paragraphs), ask the user whether to save it as a synthesis page:
If yes, write it using the Synthesis Page Format from CLAUDE.md.Would you like me to save this as wiki/syntheses/<slug>.md for future reference? - Append to
wiki/log.md:## [YYYY-MM-DD] query | <short question>
Fallback behavior
If the wiki has no relevant pages:
- Tell the user "I don't see anything about this in your wiki yet."
- Suggest running
/wiki-syncto pull in any recent sessions that might cover the topic. - If the topic is unlikely to be in session history, offer to answer from general knowledge (and make that explicit: "this is general knowledge, not from your wiki").
Don't do
- Don't make up wiki pages that don't exist.
- Don't confuse general knowledge with wiki content in the answer.
- Don't over-cite — if the whole answer comes from one page, cite once at the top.
- Don't write a wall of text. The answer should be as short as the question allows.
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.

