Persona schema
The full shape of an agent persona, generated from `core/agents/schema.py`.
Top-level fields
| Field | Type | Purpose |
|---|---|---|
id | string | Stable persona identifier |
slug | string | URL-friendly slug |
name | string | Display name |
role | reasoning | coding | general | Routes prompts |
category | string | Specialisation bucket |
preferred_model | string | null | Pin a model for all this persona's requests |
persona | PersonaConfig | Avatar, tone, language, tagline |
intelligence | AgentIntelligence | System prompt, domain rules, examples |
knowledge | AgentKnowledge | Linked documents and references |
memory | MemoryConfig | Four-layer learning state |
learning | AgentLearning | Feedback history and quality rollups |
benchmarks | AgentBenchmarks | When last benchmarked and scores |
shortcuts | list of Shortcut | Slash-command triggers |
simezu | SimezuConfig | Ecosystem integration metadata |
stats | AgentStats | Created, updated, invocation counts |
MemoryConfig (the four-layer learning system)
style: hints and preferences about phrasingstack_fingerprint: detected and confirmed tech stackdecision_patterns: how the persona has resolved trade-offs, with decayknowledge_links: files the persona auto-injects into context
Run the generator (
python scripts/export_docs_to_marketing.py) to refresh this page with every field auto-extracted from schema.py.