How do I install the Wiki Query skill?

Run npx skillmds@latest add ihudak/wiki-query in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Wiki Query safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Wiki Query?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Wiki Query free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Wiki Query?

ihudak (@ihudak) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Wiki Query by ihudak — Agent Skill

Name: Wiki Query
Author: ihudak

Wiki Query

Answer a question using the compiled wiki, with page citations and optional links back to original source files. Uses qmd MCP for semantic search when available (Claude Code), falls back to index-guided reading otherwise (Copilot). Offers to save substantive answers as new wiki pages. Triggers on: wiki-query, ask the wiki, what does the wiki say about, query the wiki, wiki query, search the wiki.

ihudak c0235fb 3.6 KB Updated Sep 17, 2026

File contents

wiki-query

Read ~/.copilot/installed-plugins/ihudak-copilot-plugins/obsidian-llm-wiki/skills/wiki-schema/SKILL.md fully before proceeding.

Invoke with: /wiki-query <question>

Step 1 — Resolve VAULT_PATH

Vault path: ~/obsidian_vault by default; set VAULT_PATH to override. WSL users with a Windows-side vault must set VAULT_PATH explicitly (e.g. /mnt/c/Users/<name>/obsidian_vault).

Step 2 — Load recent context

Read wiki/hot.md if it exists. If the file is absent or empty, skip silently. If it contains a directly relevant answer or strong context clues, use that to guide which pages to read.

Step 3 — Choose search path

Path A — qmd MCP (Claude Code, when qmd collection is registered)

Check if qmd has a wiki collection:

mcp__plugin_qmd_qmd__status

If the wiki collection is listed in the status output, use Path A. Otherwise use Path B.

Path A steps:

Run a hybrid search against the wiki collection:

mcp__plugin_qmd_qmd__query
searches: [{type: "lex", query: "<keywords>"}, {type: "vec", query: "<question>"}]
intent: "<question>"
minScore: 0.4

Also check if a meetings or sources collection is registered. If the question is about specific events, decisions, or people, run an additional query against those collections and merge the results.

Retrieve the top-scoring pages:

mcp__plugin_qmd_qmd__get
path: "<result filepath>"

Read the returned pages fully. Proceed to Step 4.

Path B — Index-guided reading (Copilot, or qmd not configured)

Read wiki/_index.md fully.
Scan the index summary column for entries relevant to the question.
Read up to 5 most relevant pages. Prioritise by: direct title match, then tag match, then summary match.
If 5 pages are not enough to answer confidently, read up to 3 more. Do not read the entire wiki.

Step 4 — Synthesise the answer

Write the answer in declarative present tense. Cite every claim:

For wiki pages: (→ [[concepts/MCP Server]])
For original source files: (→ [[Meetings/MGD/2026-04-10 MCP Planning.md]])
For multiple sources: list all

If the wiki contains contradicting information on the question: surface both sides and note the conflict. Do not silently pick one.

If the wiki does not contain enough information to answer: say so explicitly. Do not fabricate or extrapolate beyond what is in the wiki.

Step 5 — Offer to save

If the answer is substantive (involves synthesis across multiple pages, or produced a non-obvious conclusion), offer:

This answer synthesises content from N wiki pages. Save it as a new wiki page?
Suggested type: decisions/ | concepts/ | patterns/
Suggested title: "<derived from question>"
[yes / no]

If the user says yes, follow the wiki-save skill workflow.

qmd Setup Note

For enhanced semantic search, register the wiki collection once:

qmd collection add wiki --name wiki
qmd collection add Meetings --name meetings
qmd embed

The skill auto-detects whether the wiki collection is registered on each query. Without it, Path B (index-guided reading) is used — slower but always available.

---name: wiki-querydescription: Answer a question using the compiled wiki, with page citations and optional links back to original source files. Uses qmd MCP for semantic search when available (Claude Code), falls back to index-guided reading otherwise (Copilot). Offers to save substantive answers as new wiki pages. Triggers on: wiki-query, ask the wiki, what does the wiki say about, query the wiki, wiki query, search the wiki.---# wiki-queryRead `~/.copilot/installed-plugins/ihudak-copilot-plugins/obsidian-llm-wiki/skills/wiki-schema/SKILL.md` fully before proceeding.Invoke with: `/wiki-query <question>`---## Step 1 — Resolve VAULT_PATHVault path: `~/obsidian_vault` by default; set `VAULT_PATH` to override. WSL users with a Windows-side vault must set `VAULT_PATH` explicitly (e.g. `/mnt/c/Users/<name>/obsidian_vault`).---## Step 2 — Load recent contextRead `wiki/hot.md` if it exists. If the file is absent or empty, skip silently.If it contains a directly relevant answer or strong context clues, use that to guide whichpages to read.---## Step 3 — Choose search path### Path A — qmd MCP (Claude Code, when qmd collection is registered)Check if qmd has a wiki collection:```mcp__plugin_qmd_qmd__status```If the `wiki` collection is listed in the status output, use Path A. Otherwise use Path B.**Path A steps:**1. Run a hybrid search against the wiki collection: ``` mcp__plugin_qmd_qmd__query searches: [{type: "lex", query: "<keywords>"}, {type: "vec", query: "<question>"}] intent: "<question>" minScore: 0.4 ```2. Also check if a `meetings` or `sources` collection is registered. If the question is about specific events, decisions, or people, run an additional query against those collections and merge the results.3. Retrieve the top-scoring pages: ``` mcp__plugin_qmd_qmd__get path: "<result filepath>" ```4. Read the returned pages fully. Proceed to Step 4.### Path B — Index-guided reading (Copilot, or qmd not configured)1. Read `wiki/_index.md` fully.2. Scan the index summary column for entries relevant to the question.3. Read up to 5 most relevant pages. Prioritise by: direct title match, then tag match, then summary match.4. If 5 pages are not enough to answer confidently, read up to 3 more. Do not read the entire wiki.---## Step 4 — Synthesise the answerWrite the answer in declarative present tense. Cite every claim:- For wiki pages: `(→ [[concepts/MCP Server]])`- For original source files: `(→ [[Meetings/MGD/2026-04-10 MCP Planning.md]])`- For multiple sources: list allIf the wiki contains contradicting information on the question: surface both sides andnote the conflict. Do not silently pick one.If the wiki does not contain enough information to answer: say so explicitly.Do not fabricate or extrapolate beyond what is in the wiki.---## Step 5 — Offer to saveIf the answer is substantive (involves synthesis across multiple pages, or produced anon-obvious conclusion), offer:```This answer synthesises content from N wiki pages. Save it as a new wiki page?Suggested type: decisions/ | concepts/ | patterns/Suggested title: "<derived from question>"[yes / no]```If the user says yes, follow the `wiki-save` skill workflow.---## qmd Setup NoteFor enhanced semantic search, register the wiki collection once:```bashqmd collection add wiki --name wikiqmd collection add Meetings --name meetingsqmd embed```The skill auto-detects whether the `wiki` collection is registered on each query.Without it, Path B (index-guided reading) is used — slower but always available.

Frequently asked questions