radical-candor
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEMETADATA_POISONINGINDIRECT_PROMPT_INJECTION
Full Analysis
- [METADATA_POISONING]: The skill contains inconsistent and potentially copy-pasted metadata. The author is identified as 'ClawFu' in the YAML frontmatter and 'MKTG Skills' in the bottom metadata block. Furthermore, the 'Skill Boundaries' section incorrectly describes the skill's capabilities as related to 'audio production workflows' and 'audio engineering,' which is unrelated to the primary content of the skill (management feedback). These are documentation errors rather than malicious intent.
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to process user-supplied input describing interpersonal conflicts and feedback scenarios (e.g., 'Context: [relationship, situation]'). This represents an ingestion surface for untrusted data. However, since the skill provides no automated tool capabilities for file system access, network communication, or shell execution, this surface does not lead to any exploitable outcomes.
- [SAFE]: No obfuscation, remote code execution, persistence mechanisms, or credential exfiltration attempts were found in the instructions or metadata.
Audit Metadata