| name | configure-tune |
| description | Add and shape a tune section in a Plugboard YAML configuration. |
Add a tune section to a Plugboard YAML config
Use this skill when
- the user wants to tune a model
- the user wants certain parameters to become configurable or optimisable
Goal
Work with the user to decide what should be tunable, then add an appropriate tune section to the YAML config.
Instructions
- Start by asking questions to understand:
- which outcomes matter
- which component parameters are candidates for tuning
- valid ranges, choices, or distributions for each parameter
- any constraints between parameters
- Make sure the parameters being tuned are serialisable and already represented cleanly in the YAML config.
- If the current model structure hides tunable behavior inside a monolithic component, suggest splitting the logic into clearer components before or alongside tuning.
- Add a
tune section that reflects the agreed tunable parameters and their search spaces.
- Validate the updated YAML against
plugboard_schemas.ConfigSpec, since that schema includes both the process configuration and the optional tune section.
- Keep the rest of the config readable and preserve stable component names so the tuning setup remains understandable.
- Summarise which questions were answered, which defaults were assumed, how the config was validated, and what the user may still need to refine.
Output
- an updated YAML config with a
tune section
- a concise summary of the tuning choices and any open questions