Team Knowledge

Structured knowledge layer for agent teams. Three primitives: lessons, decisions, and tasks.

Agent teams accumulate experience across dozens of sessions. Without a structured way to capture it, lessons get lost, decisions get contradicted, and the same mistakes happen twice. Each agent's MEMORY.md only knows what that agent experienced.

What this skill gives your team:

• Lessons: What we learned (prevents repeated mistakes)
• Decisions: Why we chose X (prevents contradictory choices)
• Tasks: What needs doing (self-managed work queue)

All stored as plain markdown with YAML frontmatter. Searchable via grep or semantic memory (QMD). Any agent can read/write. Includes templates, maintenance protocol, and search patterns.