para-second-brain

Pass

Audited by Gen Agent Trust Hub on Feb 19, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION] (SAFE): The skill runs a local bash script to audit PARA directory structures. The script is well-contained, using standard shell utilities without network access or excessive permissions.
  • [PROMPT_INJECTION] (LOW): An indirect prompt injection surface exists via untrusted file system metadata. 1. Ingestion points: Filenames and directory names are read from the target path by scripts/validate.sh. 2. Boundary markers: None; structural metadata is interpolated into the validation report without delimiters. 3. Capability inventory: Local script execution and report generation. 4. Sanitization: None; filenames are treated as trusted strings in the report generation process.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 19, 2026, 09:20 PM