What problem does it solve? Spawning agents with the same model, skill, and behavior settings repeatedly means re-writing config every time. This Skill lets you store those settings as named, reusable presets that seed an agent's config at spawn time. ## Core Features & Use Cases - Preset CRUD: Create, list, view, update, and delete named config templates through the REST API (/api/presets) or the ava presets CLI. - Config design guidance: Explains per-agent config fields like llm_model, skills_to_expand_at_start, and skills_to_inject_into_system_prompt, including fork rules and override semantics with config_overlay. - Role card authoring: A reference guide explains how to build a role (e.g., product manager) as a be-a-<role> skill and wire it into a preset. - Use Case: You need a reusable "reviewer" agent type. Create a preset naming the model and preloaded skills, then spawn agents with preset="reviewer" instead of passing full config each time. ## Quick Start Ask the agent to create a new preset named my-preset with a chosen model and preloaded skills, then verify it with ava presets ls.
