party_get_prefetch
Retrieve pre-generated narrative variants for combat turns to accelerate storytelling. Uses cached content when available to reduce main-model calls, providing instant refined narratives for D&D campaigns.
Instructions
Retrieve a pre-generated narrative variant for a combat turn.
If the prefetch engine has a cached variant for this turn, returns a refined narrative instantly (no main-model call needed). On cache miss, falls back to full generation with the main model.
Call this right after party_thinking, before writing your own narrative. If 'cached' is true in the response, use 'narrative' as your starting point and adjust only the details that differ from actual game state.
Input Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| turn_id | Yes | Turn identifier — use the same format as the observer: 'round_{N}_{character_name}', e.g. 'round_3_Aria' | |
| outcome | Yes | Actual combat outcome: 'hit', 'miss', or 'critical' | |
| roll | No | The actual attack roll value | |
| damage | No | Damage dealt (for hit/critical) | |
| target_hp | No | Target's remaining HP after damage |