The Agent Skills Directory

[SAFE]: The skill serves as an instructional guide for prompt engineering. It defines a structured methodology (MCSLA) and provides templates for users and agents to follow to produce better video generation results.
[SAFE]: References to local project files (e.g., ../../specs/model-specs.yaml, scripts/build_index.py) and platform-specific tools (e.g., higgsfield CLI, MCP models_explore, and the ratio ledger script) are consistent with its purpose as a technical manual for a specific AI development environment.
[SAFE]: The instructions regarding 'Age-blind' prompting, 'Generic-Emotion Decomposition', and 'Conflict Resolution' are technical strategies for maximizing video generation quality and adherence to user intent within the Higgsfield ecosystem, rather than attempts to bypass agent-level safety filters.
[SAFE]: No evidence of prompt injection, obfuscation, data exfiltration, or persistence mechanisms was found during the analysis.

higgsfield-prompt