What problem does it solve? Repeatedly hitting the same Higgsfield generation failures — content filter blocks, character drift, ignored style keywords, model mismatches — wastes credits and time. This Skill checks a persistent memory of past failures and confirmed fixes before any prompt is written, so known problems are pre-empted instead of rediscovered. ## Core Features & Use Cases - Pre-Generation Recall: Automatically queries filter-memory and quality-memory databases with key terms extracted from the intended prompt, then silently applies confirmed substitutions and fixes. - Generation Ledger Logging: Records every generation outcome (kept or rejected) with controlled-vocabulary reason tags, enabling takes-per-kept ratios and iterate-vs-batch verdicts per shot type. - Routing Telemetry: Logs which sub-skills were used per request so teams can see which skills are load-bearing over time. - Use Case: Before writing a prompt for a two-character dialogue close-up, the Skill finds that a previous attempt with the same model produced lip-sync desync and that a confirmed fix exists — it applies that fix to the new prompt without the user needing to remember the failure. ## Quick Start Ask the agent to write a Higgsfield prompt for your scene and it will automatically check past failure memory and apply known fixes before presenting the prompt.
