team-brain

An LLM-native team wiki and agent-driven development workflow — in one template.

Team-Brain is a reusable knowledge system for teams working across multiple repositories. It is the synthesis layer: repo-local docs stay authoritative for implementation details, while this repo preserves how the product concept, architecture, decisions, operations, and open questions fit together — and ships the agent skills that run a full idea-to-wiki development loop on top of it.

The one rule: wiki/ contains only decided and implemented knowledge. Everything pre-implementation lives in inbox/ and plans/.

⚡ Quick Start

• Clone the template (or use it as a GitHub template repo) next to your team's repos:

• Open it in your agent tool. The skills ship pre-installed for Claude Code (.claude/skills/), Cursor (.cursor/skills/), and Codex (.codex/skills/) as mirrors of the canonical .agents/skills/ — no install step needed.

• **Run /setup-brain.** It detects whether you're configuring a fresh template or migrating an existing wiki, walks the customization checklist (repos.yaml, namespaces, zones, placeholders), prunes the skill mirrors for tools you don't use, and hands off to /skills-sync for installing skills into your implementation repos. Re-running it later doubles as a setup health check.