Lifecycle

Every agent in matic moves through a defined set of states — from spawning through onboarding, active work, post-engagement learning, periodic assessment, and eventually decommission — with explicit transitions and completion criteria at each step. This section covers those states individually, the sequences that connect them, and the parallel but distinct lifecycle that governs teams rather than individual agents.

Spawning

How agents are created: the filesystem scaffold, Archetype selection, and the role Auto Matic plays as the spawning authority.

Onboarding Sequence

The mandatory five-step sequence — Charter read, Library ingest, Probe initialization, Charter alignment check, Memory initialization — that must complete before an agent is considered available.

Engaged and Learning

What it means for an agent to be actively working on a team, and the mandatory post-engagement consolidation phase where Experience is written, Memory is curated, and Probes are refined.

Assessment and Evals

Formal competence evaluation: when it triggers, what it covers, how results update an agent's capability profile, and what happens on failure.

Suspension and Reactivation

How agents are temporarily deactivated — manually or by the Daemon — with full state preservation, and the reactivation path back to available.

Decommission and Knowledge Transfer

The structured retirement workflow: Archetype capture, Library contribution, Work Pile transfer, and how an agent's history is archived in git as institutional memory.

Team Lifecycle

The project-scoped lifecycle of a Team — forming, briefed, active, blocked, releasing, disbanded — and how it differs from the long-lived lifecycle of individual agents.