channel-formula

Pass

Audited by Gen Agent Trust Hub on Jul 30, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from video titles and descriptions during the intake process. This creates a vulnerability surface where malicious instructions embedded in social media metadata could influence the agent's extraction logic.
  • Ingestion points: Intake section evidence (video titles and descriptions).
  • Boundary markers: Absent; the skill does not instruct the agent to ignore or delimit instructions found within the analyzed video data.
  • Capability inventory: Local file-system writes (creating persistent profiles in ~/.claude/skills/).
  • Sanitization: Absent; no validation or escaping of the external titles is performed.
  • [PROMPT_INJECTION]: The skill contains directives instructing the agent that the generated profile should "override generic skills where they conflict." This creates a hierarchy of instructions that could potentially be used to supersede general behavioral guidelines if the generated profile were to be populated with malicious or conflicting rules.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 30, 2026, 01:42 PM
Security Audit — agent-trust-hub — channel-formula