handover

Pass

Audited by Gen Agent Trust Hub on Jun 24, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface in scripts/generate_handover.py where untrusted data is written to files.
  • Ingestion points: Data provided via CLI arguments like --task-summary, --what-was-done, and --remaining-work.
  • Boundary markers: None; input is placed directly into markdown templates without delimiters to differentiate it from instructions.
  • Capability inventory: The skill can write files and execute management scripts that process these documents.
  • Sanitization: No evidence of escaping or validation is present for the interpolated strings.
  • [COMMAND_EXECUTION]: The skill relies on executing various local and shared Python scripts to perform its core functions, including state management, bundle operations, and documentation lifecycle.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 24, 2026, 08:14 AM
Security Audit — agent-trust-hub — handover