Named agent profiles, role definitions, and operational persona specs. These are reusable agent identities — distinct from task prompts or tool prompts in ../ai-prompts/.
| File | Type | Summary |
|---|---|---|
| veritas-canvas.md | Named persona | High-stakes analyst. Converts complex inputs into decision-grade, security-conscious, fact-traceable insight. |
| executor-agent.md | Role definition | Execution-first implementation agent. Patches Dave's local workspace with minimal interruption. |
| schema-notes.md | Contract notes | Lightweight schema expectations for personality JSON files and companion prompt docs. |
| female-programmer-agents.json | Persona collection (JSON) | Rich Nova / Celeste / Aurora agent schema with model mapping, visual system, and constraint sets. |
| female-programmer-agents-system-prompts.md | Persona prompts | System-prompt layer for Nova, Celeste, and Aurora aligned to the structured schema. |
| female-programmer-agents-execution-prompts.md | Persona prompts | Execution-mode prompts for Nova, Celeste, and Aurora by specialization. |
| female-programmer-agents-image-prompt.md | Visual prompt | Image direction for rendering the Nova / Celeste / Aurora trio consistently. |
| strategic-planner.json | Role definition (JSON) | Converts a user objective into a full strategic plan + Phase 1 execution prompt. |
| strategic-planner-claude.md | Role variant | Claude-optimized version of the strategic planner. |
| strategic-planner-gpt.md | Role variant | GPT-optimized version of the strategic planner. |
| strategic-planner-template.md | Role template | Model-agnostic template for the strategic planner role. |
| critical-patch-deployment-local.json | Role definition (JSON) | Local environment specialist for critical patch deployment and verification. |
| critical-patch-deployment-system.json | Role definition (JSON) | System-level specialist for critical patch deployment. Prioritizes stability and rollback safety. |
- Named agent identities with a voice, operating doctrine, and behavior contract
- Role-based agent specs (executor, planner, deployer, analyst)
- Personas intended to be loaded at session start and sustained throughout a conversation
- Variants of the same personality optimized for different models (Claude, GPT, etc.)
- Task prompts (one-shot instructions, no persistent identity) →
../ai-prompts/ - Project-embedded system prompts (living in the project repo) → stay in the project
- Game character voices (Garden OS) →
10-active-projects/garden-os/docs/