| name | knows |
| description | Unified knowledge retrieval skill. Build traceable answers from git, MCP, and memory files. |
| version | 2.0.0 |
| last_updated | "2026-05-28T00:00:00.000Z" |
Knows
Answer with evidence using sources in strict order: Git → MCP (mind/graph) → Memory files
When To Use
- Why behavior/logic changed (git blame + commit history)
- Impact radius of function/screen/API (graph_mcp traversal)
- Architecture/decisions context (mind_mcp + memory files)
- NOT for: quick syntax fixes, pure implementation, DB mutations
Evidence Priority
| Question Type | Priority Order |
|---|
| Structure/Runtime | graph_mcp → Git → mind_mcp → memory |
| Historical/Rationale | mind_mcp → memory → Git → graph_mcp |
Workflow
Phase 1 - Preflight: Classify intent (why-changed, impact-analysis, architecture-context, history-trace)
Phase 2 - Git:
rg -n "keyword" .
git log --oneline --decorate -- <path>
git blame -L <start>,<end> <file>
git show <commit>
Phase 3 - MCP:
mind_mcp: hybrid_search, query_graph_rag_relation, sequential_search
graph_mcp: explore_graph, query_subgraph, find_paths, analyze_workflow_impact
Phase 4 - Memory: Workspace-first, then home. Allowlist: memory*.md, agent*.md, claude*.md, cursor*.md
Phase 5 - Synthesis: Merge by priority. Separate facts/inferences/conflicts.
Output Format
## Kết luận ngắn
## Độ tin cậy
- Confidence: high/medium/low
- Nguồn: list sources used
## Bằng chứng
- Git: file, commit, author
- MCP: query type + summary
- Memory: file + excerpt
## Điểm chưa chắc chắn
## Neo4j Query Suggestion (optional)
Guardrails
- Never claim from single weak source
- Every high-impact claim needs citation
- Do not execute direct Neo4j queries
- Redact secrets before citing
Limits
- MCP timeout: 30s/call, 5min total
- File size: 300KB max
- Files/query: 10 max
- Cache TTL: 10min
Degraded Mode
⚠️ Degraded Mode: {reason}
- Unavailable: {failed channels}
- Missing: {limited evidence}
- Confidence: {downgraded reason}