| name | query |
| description | Answer questions about the wiki by searching, reading, and synthesizing relevant scraps with `[[Title]]` citations. Use this when the user wants to query the wiki, ask what they have written about a topic, compare scraps, find related notes, or pull a synthesized overview from existing scraps. |
| allowed-tools | Read, Bash, Glob, Grep |
| user-invocable | true |
| argument-hint | ["question"] |
Query
Answer a question against the wiki and return a citation-rich synthesis.
Implements Karpathy's Query primitive for Scraps: search the wiki, read relevant scraps, and synthesize an answer that references its sources by [[Title]]. Citations make the answer auditable and ready to be filed back as a new scrap if the user chooses.
When to use
- "What do I have on X?" / "What did I write about Y?"
- "Compare X and Y across my notes"
- "Summarize what I know about Z"
- "Find scraps related to W"
Workflow
-
Understand the question
- Identify the core topic and any constraints (time range, ctx folder, tag, etc.)
- If the question is ambiguous, ask one clarifying question before searching
-
Search (broad → narrow)
scraps search "<keyword>" --json for each main keyword- Try multiple phrasings if the first search returns few results
- For tag-driven questions:
scraps tag backlinks "<tag>" --json
-
Select candidates
- From search results, pick the 5–15 most relevant scraps
- Prefer scraps that span the question (different ctx, different tags) over many near-duplicates
-
Read selected scraps
scraps get "<title>" [--ctx <ctx>] --json for each candidate- Use field projection to save context when appropriate, e.g.
--json body, --json headings, or --json code_blocks
- If a link result includes
heading, read just that section with scraps get "<title>" [--ctx <ctx>] --heading "<heading>" --json body
- For graph-shaped questions, also use:
scraps links "<title>" --json (outbound)scraps backlinks "<title>" --json (inbound)
-
Synthesize with citations
- Write the answer in plain Markdown
- Cite every claim that comes from a scrap as
[[Title]] (or [[Ctx/Title]] when needed)
- Do not invent information beyond what the scraps and the user's question support
- Choose the output shape that fits the question:
- prose paragraph for narrative questions
- GFM table for comparisons
- bullet list for enumerations
- mermaid diagram for relationships
-
Stop
- Do not auto-offer to save the answer. If the user wants the synthesis saved as a new scrap, they invoke the
ingest skill on the answer markdown.
Citation rules
- Every non-trivial claim should trace to a
[[Title]] citation.
- If a claim has no source in the wiki, mark it as outside the wiki ("not in the current scraps"); do not bluff.
- Use the exact
title value returned by scraps search --json / scraps get --json so the citation resolves cleanly.
- For ctx-disambiguated scraps, use
[[Ctx/Title]].
scraps links --json may return a heading; use it for targeted reads, but cite the scrap as [[Title]] or [[Ctx/Title]] unless the answer specifically needs a heading reference.
CLI used
scraps search <query> [--logic and|or] --jsonscraps get <title> [--ctx <ctx>] [--heading <heading>] --json [fields]scraps links <title> [--ctx <ctx>] --json (returns { kind, title, ctx, heading } refs)scraps backlinks <title> [--ctx <ctx>] --jsonscraps tag list --jsonscraps tag backlinks <tag> --jsonscraps todo [--status open|done|deferred|all] --json
These commands are shown with --json because this skill consumes structured output — without it Scraps prints human-formatted text that is unstable to parse, so --json is part of the command, not an option. Use scraps <cmd> --help only as a fallback for a rare flag not listed here, not as a routine discovery step; for deeper syntax or spec questions, consult the scraps-llm-wiki-schema agent (it reads the official docs).
Further reading
