fusion-skill-authoring
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides a robust framework for authoring other agent skills, focusing on clear metadata and discovery cues.
- [SAFE]: It includes explicit safety constraints in the instructions, forbidding hardcoded secrets, remote code execution, and hidden network access.
- [SAFE]: A provenance check mechanism is included to protect against accidental modifications of installed dependency files.
- [SAFE]: The instruction set promotes progressive disclosure by offloading complex examples and deep guidance to specialized reference and asset files.
- [SAFE]: Helper agents are strictly limited to scoping, review, and trigger tuning roles, minimizing the risk of autonomous unauthorized actions.
Audit Metadata