delegation-mode

Pass

Audited by Gen Agent Trust Hub on Aug 15, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill implements a workflow where the agent resumes work by reading state from files in the .agents/tasks/ directory. This creates a surface for indirect prompt injection where malicious instructions embedded in these files could potentially influence agent behavior.
  • Ingestion points: The agent is instructed to read request.md, status.md, and result.md from the .agents/tasks/ directory during worker procedures and when resuming work.
  • Boundary markers: The instructions do not define delimiters or warnings to isolate or ignore instructions within these state files.
  • Capability inventory: The skill utilizes file system read and write operations to manage task lifecycle states (PENDING, RUNNING, PAUSED, BLOCKED, COMPLETE).
  • Sanitization: No validation or sanitization processes are specified for the content of the ingested files.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 15, 2026, 08:59 AM
Security Audit — agent-trust-hub — delegation-mode