Archetypes
An archetype is a reusable blueprint that defines the character, capabilities, and behavioral profile of an agent class — not an agent itself. When matic spawns an agent from an archetype, it loads a fixed persona, a set of skills, and an initial probe profile; from that point forward the agent develops independently through accumulated experience. Archetypes are versioned, capturable from experienced agents, and depreciable — making them the primary mechanism for institutionalizing organizational knowledge about how agents should operate.
Archetype Lifecycle
How an archetype moves from initial human authorship through agent spawning, experience accumulation, and eventual capture or retirement.
Archetype Capture
The three-step protocol — CLI proposal, Auto Matic rubric review, human PR approval — for distilling a proven agent into a new or updated archetype.
Built-In Archetypes
The archetypes matic ships by default, covering common organizational roles such as researcher, engineer, reviewer, and coordinator.
Persona
The identity, communication style, and cognitive disposition defined by an archetype and fixed at agent spawn time.
Skills
Knowledge documents that give tools operational context — prescribing how agents use capabilities, what sources are authoritative, and how to structure execution for a given category of work.
Probe Schemas
The schema and DSL for capability, confidence, constraint, preference, and trigger probes that archetypes seed into every spawned agent.
Versioning and Deprecation
How archetype versions are numbered, when a version is deprecated, and the upgrade path available to agents still running on older definitions.