The one-screen card for the 📝Agentic Collaboration System (ACS) — every operational artifact in 📝human–AI collaboration, sorted by what it does. Four planes answer four questions: does it say what to do (Logic), trigger something (Binding), track what's happening (State), or orient a run to its terrain (Context)?
The Classes, by Verb
Logic — what to do
- Policy — Prescribes. Standing rules, always in force.
- Procedure — Runs on demand. Ordered steps you hand work to.
- Loop — Runs on a timer. Steps an agent self-triggers, finding its own work.
Binding — how it's triggered
- Command — Triggers by hand. A human-invoked pointer to Logic.
- Hook — Triggers on an event. Fired automatically by the harness.
- Script — Runs code. Deterministic, no agent judgment.
- Skill — Loads capability. An instruction/code bundle, on demand.
State — what a run reads and writes
- Roadmap — Tracks work. Items through stages, per domain.
- Initiative — Tracks a scoped push. Born to die; harvested on completion.
- Report — Records a run. One daily-journal bullet appended at close-out (OSR memos retired 2026-07-01).
Context — what a run reads to orient
- Context — Describes. The operating manual, read to orient — never prescribes (that's Policy), triggers, or tracks.
The Parent Chain
Product memo → Stewardship Context → { Roadmap · Policy · Loops · … }. # Parent is the universal up-link every artifact declares; a project's tree roots at its product/entity memo.
Tags
Three families under each namespace: membership #<ns> · classification #<ns>-<class> · status #<ns>-active / -dormant / -deprecated. They compose — an active loop is #<ns>-loop + #<ns>-active, so a two-tag search lands straight on it.