analytics
Pass
Audited by Gen Agent Trust Hub on Sep 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to read local context files (such as .agents/product-marketing.md) to gather background information. This creates a surface for indirect prompt injection if those files contain malicious instructions, as the skill does not specify boundary markers or sanitization for this ingested data.
- Ingestion points: The Initial Assessment section in SKILL.md directs the agent to read context files before asking questions.
- Boundary markers: There are no instructions for the agent to treat the content of these files as untrusted or to use delimiters to prevent instruction override.
- Capability inventory: The skill's primary function is instructional, focusing on generating tracking plans and client-side code snippets rather than executing system-level commands.
- Sanitization: No validation or filtering is performed on the content of the marketing context files.
- [EXTERNAL_DOWNLOADS]: The skill provides documentation and implementation examples that reference standard tracking scripts from well-known analytics services.
- Evidence: SKILL.md and the implementation references mention gtag.js and the Facebook Pixel (connect.facebook.net), which are industry-standard tools for web measurement and are considered safe references.
Audit Metadata