quick-capture
Pass
Audited by Gen Agent Trust Hub on Apr 29, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill ingests untrusted text and audio data that influences its classification and routing logic.
- Ingestion points: User-supplied text and audio paths provided via $ARGUMENTS in SKILL.md.
- Boundary markers: The skill uses YAML frontmatter in the generated inbox files to separate metadata from content but lacks explicit instructions to the processing agent to ignore instructions embedded within the verbatim captured text.
- Capability inventory: The skill has the capability to write to the file system and invoke the route-capture skill for potential side effects based on the captured content (SKILL.md).
- Sanitization: No sanitization is performed, as the skill is intended to capture user notes verbatim.
Audit Metadata