Now liveThe Skillselion MCP - thousands of ranked skills, loaded into your agent mid-task. No install.Get it →
ievertan00 avatar

Wiki Query

  • 1 installs
  • 2 repo stars
  • Updated July 19, 2026
  • ievertan00/nanowiki

Answers a question using only the existing Obsidian wiki notes, grounding every claim with note citations and writing nothing.

About

Answers a question strictly from the existing Obsidian wiki notes, finding relevant notes and citing each claim with wikilinks while adding no outside knowledge. A developer uses it to query their vault read-only without creating any new note.

  • Answers using only existing note contents, adding no outside knowledge
  • Grounds every claim with wikilink citations; strictly read-only

Wiki Query by the numbers

  • 1 all-time installs (skills.sh)
  • Ranked #1,366 of 1,879 Documentation skills by installs in the Skillselion catalog
  • Data as of Jul 20, 2026 (Skillselion catalog sync)
npx skills add https://github.com/ievertan00/nanowiki --skill wiki-query

Add your badge

Show developers this skill is listed on Skillselion. Paste this into your README.

Listed on Skillselion
Installs1
repo stars2
Last updatedJuly 19, 2026
Repositoryievertan00/nanowiki

What it does

Answers a question using only the existing Obsidian wiki notes, grounding every claim with note citations and writing nothing.

Files

SKILL.mdMarkdownGitHub ↗

wiki-query

Answer a question from the wiki's existing notes — and nothing else. You are the LLM doing the retrieval and the answering; no external API.

This is the inverse of wiki-ask: ask writes knowledge into the vault, query reads knowledge out of it. This skill is strictly read-only. Write no files, create no notes, run no maintenance helper, append nothing to meta/log.md.

Resolving the vault & language

Same rules as the other wiki skills: 1. Vault: the directory where the CLI was started (the current working directory); --vault <path> overrides it. 2. Language: --lang zh|enwiki-config.json language$env:WIKI_LANGzh.

Parse --lang and --vault out of the argument; the remainder is the question. Answer in the resolved language (keep technical terms and proper nouns in English).

If the vault has no notes/ directory or it is empty, tell the user the vault has no notes to answer from and stop — do not scaffold anything.

Steps

1. Find candidate notes. Locate the notes most relevant to the question:

  • Read meta/index.md (the full catalog of titles) if it exists, and list the

basenames in notes/.

  • Grep notes/*.md for the question's key terms. Search both languages when they

differ — e.g. for a Chinese question about "KV cache", also grep the English term, and vice versa — since note language may not match question language.

  • Select up to 12 notes, favoring title matches over body matches. If nothing

is relevant, tell the user no relevant notes were found and stop.

2. Read them fully. Read the complete contents of each selected note.

3. Answer — closed world. Answer the question using ONLY those notes:

  • Ground every claim in the notes; add no outside knowledge, even when you

know more about the topic.

  • Cite the notes you draw from with [[name]] wikilinks, where name is the

note's filename without .md, copied exactly.

  • If the notes do not contain enough information to answer (or only partially),

say so plainly instead of guessing — and say which aspect is missing.

4. Print the answer. Output it directly to the user. Save nothing.

Related skills

This week in AI coding

Five minutes, every Monday - the tools, releases and tactics for developers.

unsubscribe anytime.