travel-planning
Pass
Audited by Gen Agent Trust Hub on Jul 13, 2026
Risk Level: SAFEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The
README.mdfile exposes environmental metadata by including absolute Windows file paths (e.g.,C:/Users/mick0/OneDrive/Documents/Code%20&%20Dev/GitHub/AI-Skills/claude/skills/travel-planning/SKILL.md). This reveals the author's local username and directory structure to the AI agent. - [PROMPT_INJECTION]: The skill possesses an Indirect Prompt Injection surface due to how it handles local reference data.
- Ingestion points: The skill reads external data from
references/europe-2026.mdandreferences/colorado-2026.mdto inform its planning logic. - Boundary markers: Absent. The instructions do not use delimiters or provide the agent with guidance to ignore potential instructions embedded within these reference files.
- Capability inventory: The skill is designed for text-based planning and budget estimation and does not request access to shell, file-system writing, or network tools.
- Sanitization: No sanitization or input validation is applied to the content of the reference files before they are processed by the agent.
Audit Metadata