managing-timelines

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
  • [PROMPT_INJECTION] (LOW): An indirect prompt injection surface exists in the project intake phase where the skill processes untrusted user data. Evidence Chain: 1. Ingestion points: SKILL.md (Step 1) and references/INTAKE.md. 2. Boundary markers: Absent. 3. Capability inventory: No tools or dangerous capabilities (no shell, network, or file operations). 4. Sanitization: Absent. The risk is minimized by the lack of actionable tools.\n- [DATA_EXFILTRATION] (SAFE): No network operations, hardcoded credentials, or access to sensitive local file paths (e.g., ~/.ssh, ~/.aws) were detected.\n- [REMOTE_CODE_EXECUTION] (SAFE): No patterns for remote script execution (curl|bash), dynamic code evaluation (eval/exec), or external package installation were found.\n- [NO_CODE] (SAFE): This skill is entirely instructional and consists of Markdown documentation and templates. No executable code or scripts are included in the package.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:25 PM