Should TeamAI support shareable team workflows?

[jeff-r2026]

Anthropic's AI-Native SDLC Playbook describes a workflow where each stage ends by committing an artifact (intent.md → spec.md → plan.md → code → PR), and the next stage begins by reading it. Standards are encoded as skills, and humans review at the gates.

That's one workflow. Every team has its own, and the playbook itself says the stages are modular.

TeamAI already syncs the pieces a workflow is made of:

• Rules — when the workflow applies (an always-on rule reaches every session)
• Skills — how each stage is done
• Hooks — gates that block the next step
• Docs — artifacts carried between stages

What's missing is that nothing ties them together. After teamai pull, a member gets a flat list of skills, and neither they nor the Agent knows those files are one workflow, or which stage comes next.

So: should a workflow be a first-class concept in TeamAI, or is a convention plus a forkable skeleton in the team repo enough?

My leaning is the skeleton first — let a few teams adapt it to their own stages, and add a workflow manifest only if they hit real problems with discovery or enforcement. No new CLI command, no workflow engine.

Questions

1. Do you have a repeatable AI workflow worth sharing with your team?
2. Should TeamAI just distribute workflows, or also enforce them?