writing-plans

Pass

Audited by Gen Agent Trust Hub on Mar 6, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection through its data ingestion process.
  • Ingestion points: Data is received via the bd show command and the output report from the codebase-investigator agent.
  • Boundary markers: There are no explicit delimiters or warnings used to separate external content from internal instructions.
  • Capability inventory: The skill can modify task designs via the bd update command and generate executable shell scripts.
  • Sanitization: No validation or escaping of external inputs is performed prior to interpolation into task designs.
  • [COMMAND_EXECUTION]: The skill dynamically generates and recommends the execution of shell commands and source code snippets (e.g., pytest, bd update) based on task requirements and codebase scans.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 6, 2026, 05:19 PM