codex-handoff

Pass

Audited by Gen Agent Trust Hub on Jul 28, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill defines a structured workflow for generating continuity prompts. It explicitly instructs the agent to omit secrets, credentials, and conversation transcripts, which serves as a security best practice. No patterns designed to bypass model constraints were found.
  • [DATA_EXFILTRATION]: The skill manages state transfer between platform threads. It prohibits the inclusion of sensitive data in the transfer and does not perform any external network requests or unauthorized file access.
  • [COMMAND_EXECUTION]: All actions are performed through high-level platform tools like create_thread and list_projects. The instructions specifically forbid the use of shell commands or local file creation for handoff purposes.
  • [SAFE]: The skill is designed with security-conscious constraints and clear functional boundaries, making it safe for its intended purpose of context management.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 28, 2026, 11:42 AM
Security Audit — agent-trust-hub — codex-handoff