| name | wiki-query |
| description | Query the knowledge graph for answers. Searches knowledge entities, claims, sources, and cross-references with code. Cites specific graph nodes, not training data. Triggers on: "what do we know about", "query knowledge", "check the wiki", "knowledge search".
|
| allowed-tools | mcp__codescope__knowledge_search mcp__codescope__semantic_search mcp__codescope__explore mcp__codescope__find_callers mcp__codescope__impact_analysis Read |
| user-invocable | true |
| argument-hint | <question> |
wiki-query: Knowledge Graph Query
Answer the user's question using ONLY the knowledge graph. Cite specific nodes. If the knowledge graph doesn't have the answer, say so — don't fall back to training data.
Query Strategy
-
Search knowledge graph first:
knowledge_search(query: "[user's question]")
-
If knowledge results reference code, drill into it:
find_callers(function_name: "[referenced function]")
explore(entity_name: "[referenced entity]")
-
If knowledge graph is empty for this topic, say:
"No knowledge found on this topic. Run /autoresearch [topic] or /wiki-ingest [source] to add it."
Answer Format
## [Question rephrased as title]
[Answer in 2-5 paragraphs]
### Sources
- [[Entity/Concept name]] — [what it contributed to the answer]
- [[Source name]] — [key claim from this source]
### Related Code
- `function_name` (file.rs:42) — [how it relates]
### Confidence: [high/medium/low]
[Why this confidence level — how many sources, any contradictions]
Rules
- NEVER answer from training data when knowledge graph has relevant nodes
- ALWAYS cite the specific knowledge node (title + kind)
- FLAG if you find contradictions between knowledge nodes
- If answer requires both knowledge AND code, combine them
