Skip to content
Research
Command

/query

Ask read-only questions against the wiki. Supports quick, standard, deep, list, archived, supplementary-wiki, and resume modes with exact file citations.

BOOST
From plugin
nvk-llm-wiki
1.4k28 skills28 commands
Install
> /plugin marketplace add nvk/llm-wiki
> /plugin install wiki@llm-wiki

How it fires

How this command gets triggered: by you, by Claude, or both.

  • Fires itselfClaude auto-loads it when your prompt matches the work.
  • You can call itInvoke it directly when you want it.
  • Slash command/query

Context preview

What this command does when you run it.

Ask read-only questions against the wiki. Supports quick, standard, deep, list, archived, supplementary-wiki, and resume modes with exact file citations.

Command definition

query.md
description: "Ask read-only questions against the wiki. Supports quick, standard, deep, list, archived, supplementary-wiki, and resume modes with exact file citations."
argument-hint: "<question> [--quick|--deep] [--raw] [--list] [--include-archived] [--resume] [--tag <tag>] [--category concepts|topics|references] [--with <wiki>...] [--wiki <name>] [--local]"
allowed-tools: Read, Glob, Grep

Read-Only Wiki Query

Read `skills/wiki-manager/references/query-lite.md`, then answer `$ARGUMENTS` from the selected wiki. This command is always read-only: do not update indexes or append to `log.md`.

Parse

  • Everything that is not a flag is the question.
  • `--quick`: use index summaries only.
  • No depth flag: standard index-first query.
  • `--deep`: inspect all relevant compiled articles, links, and raw evidence.
  • `--raw`: allow targeted raw-source reads; implied by `--deep`.
  • `--list`: return ranked matching files instead of a synthesized answer.
  • `--include-archived`: explicitly permit archived reads and label them.
  • `--resume`: give a compact activity briefing before answering any question.
  • `--tag` and `--category`: constrain candidate selection.
  • `--with <wiki>`: use an active supplementary wiki as secondary context.
  • `--wiki <name>` and `--local`: select the primary wiki.

Depth

Quick

Read the primary `_index.md` and only the relevant branch indexes. Answer from their summaries and tags. If they are insufficient, say so and recommend the standard mode. Cite the index paths used.

Standard

Use the query-lite protocol: master index, relevant branch index, then the minimum exact articles. Use one bounded Grep only when indexes miss a likely match. Follow directly relevant See Also links. Cite exact files and surface confidence or evidence gaps that affect the answer.

Deep

Read all relevant branch indexes and articles, follow relevant cross-links, search `wiki/` and `raw/` with bounded patterns, and inspect active sibling indexes for overlap. Archived sibling indexes may be reported separately, but archived article bodies require `--include-archived`.

List Mode

Return a compact ranked list. Rank title matches above summary matches, summary above body matches, and multiple-term matches above single-term matches. Show title, exact path, summary, and tags. Include raw matches only with `--raw`. Keep archived results separate and only include them when explicitly allowed.

Resume Mode

Start with `<wiki-name> booted from <wiki-root-path>`. Read only the active session/checkpoint files, the recent tail of `log.md`, the master index stats, and the three most recently updated index entries. Report interrupted work, recent activity, stats, recent articles, and concrete next steps. If a question is present, answer it afterward using standard depth.

Supplementary and Archived Wikis

The primary wiki provides the subject. `--with` wikis provide secondary craft or domain context and must remain clearly attributed. Reject archived primary or supplementary targets unless `--include-archived` is present. Label every archived citation.

Output

Answer directly with exact file citations. For standard or deep answers, add short `Sources used` and `Knowledge gaps` sections only when useful. For list or resume-only output, omit those sections. Never use training data to fill a wiki gap and never mutate the wiki while answering.

Read more
Ships withnvk-llm-wiki

LLM-compiled knowledge bases for any AI agent. Parallel multi-agent research, thesis-driven investigation, source ingestion, wiki compilation, querying, and artifact generation.

Get the whole plugin
Stats
1,394
Stars
131
Forks
Active
Maintenance
Python
Language
MIT
License
6d ago
Last commit
6mo ago
Created
1d ago
Added

Repo: nvk/llm-wiki

Other commands on nvk-llm-wiki.