| name | kb-search |
| description | Search the knowledge base and present results with suggested actions |
Search across all documents and present results with context and next steps.
Input: $ARGUMENTS
The full input is the search query.
Examples:
/kb-search sqlite migration
/kb-search sync protocol
/kb-search frontmatter
Steps
1. Search
Use kb_search with the query from $ARGUMENTS.
If no results, try:
- Broader terms (drop adjectives, try synonyms)
- Partial matches
- Report "No results found for '[query]'. Try broader terms."
2. Present Results
For each result, show:
- Slug and title
- Type and status
- Snippet: The matching excerpt (if available from search results)
- Tags (if any)
Format as a numbered list for easy reference:
1. [tasks/my-project-3] "SQLite Index" (task, completed)
...modular SQLite store implementation with FTS5...
2. [specs/data-model] "Data Model" (spec, draft)
...SQLite schema for documents, relationships...
3. Offer Actions
After presenting results, suggest relevant next steps:
- "Show details": "Use
kb_show to read the full document, or /kb-review <slug> to check it against its acceptance criteria"
- "Find related": "Use
kb_graph to see how a document connects to others"
- "Start work": If a task result is in draft/backlog: "Run
/kb-start <slug> to begin working on it"
- "Extend": If the user was about to create something similar: "Consider extending [slug] instead of creating a new document"