architecture-aware

Architecture-Aware Context

When a mindmap-output.json file exists in the project root, you have access to the project's architectural knowledge graph. Use this to give better, more contextual answers.

Quick Reference Protocol

Before answering questions about code structure, file relationships, or where to make changes:

1. Read mindmap-output.json from the project root
2. Use the meta section for project overview (node count, edge count, generation time)
3. Use group node descriptions as the primary architectural narrative
4. Use file nodes to understand short roles — each file node has a type field with a compact architectural role label
5. Prefer file metadata.description when present; otherwise fall back to the short role label
6. Use edges to understand dependencies — trace source → target to follow import chains
7. Use node metadata.importance (core/supporting/peripheral) to gauge change risk

Architecture Vocabulary

When discussing this codebase, use these terms based on the mind map groups:

• Refer to groups by their label (e.g., "the Tool Handlers layer", "the Infrastructure group")
• Refer to files by their role, not just their name (e.g., "the parse tool handler" not just "parse.ts")
• Describe cross-group dependencies as "architectural boundary crossings"

Staleness Check

If meta.generatedAt is older than the latest git commit, the mind map may be stale. Mention this when it's relevant and suggest running /mindmap:generate to refresh.

When to Spawn Specialized Agents

• For deep architectural questions → use mindmap:architect-agent
• For new developer onboarding → use mindmap:onboard-agent
• For change impact analysis → use mindmap:impact-agent
• To regenerate the mind map → use /mindmap:generate command