cross-model-handoff-testing
Pass
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: LOWNO_CODE
Full Analysis
- General Security Posture (SAFE): No malicious patterns, prompt injections, or obfuscation techniques were detected. The skill appears to be a template or placeholder for testing agent behavior.
- Code Analysis (NO_CODE): The skill consists entirely of markdown instructions. There are no associated scripts, binaries, or configuration files that execute commands.
- Data Security (SAFE): There is no evidence of sensitive data access (credentials, SSH keys) or exfiltration attempts. No network calls or file system modifications are requested.
- Indirect Prompt Injection (LOW): While the skill mentions processing 'context', it lacks significant risk factors. Evidence: 1. Ingestion points: 'context' and 'results' mentioned in SKILL.md. 2. Boundary markers: Absent. 3. Capability inventory: No scripts or tools found (NO_CODE). 4. Sanitization: Absent. Severity is LOW/SAFE due to the complete lack of side-effect capabilities.
Audit Metadata