threat-vector-security/skills/knowledge-search

Knowledge Search

Use indexed search when the user wants answers from local docs, notes, wikis, or repository content that has already been ingested into the search service.

Workflow

1. Use doc_search for discovery across indexed collections.
2. Read only the specific files or passages you need after the search narrows the scope.
   • Use fs_read for exact follow-up reads, including direct text extraction from a known local PDF path.
3. Use doc_search_status when the user asks about indexing health or when search results look unexpectedly empty.
4. Use doc_search_reindex only when stale or missing search state is actually blocking the task.
5. When the user wants a summary of local docs or indexed material, summarize from the retrieved passages directly instead of depending on an external summary CLI unless that reviewed third-party skill has been explicitly approved and enabled.

Practical Rules

• Prefer indexed search for broad discovery.
• Prefer fs_read or fs_search only after you know the file or exact target.
• For one specific local PDF file, prefer fs_read; for many PDFs or broad discovery, index them and use doc_search.
• Tell the user when the answer is coming from indexed content versus a direct file read.

Gotchas

• Do not reindex first; confirm that stale or missing index state is actually the blocker.
• Do not present indexed-search results as if they cover files that were never ingested.
• Do not read large numbers of files after search if a few targeted passages answer the question.