gtm-context
Pass
Audited by Gen Agent Trust Hub on Oct 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest significant amounts of untrusted data from the user (such as product descriptions and customer details) and use this content to generate strategy documents, creating an attack surface for indirect prompt injection.
- Ingestion points: Phase 1 (Intake) in SKILL.md explicitly requests various text-based inputs from the user including company descriptions and customer data.
- Boundary markers: The skill does not provide instructions to the agent to use delimiters or to disregard instructions that may be embedded within the user-supplied content.
- Capability inventory: The skill instructs the agent to execute a local validation script (scripts/check-output.py) using the generated content.
- Sanitization: No evidence of sanitization, filtering, or validation of the user-provided text is present before the agent processes it.
- [COMMAND_EXECUTION]: The skill documentation directs the agent to execute a local script to verify the quality of the generated output.
- Evidence: The references/framework-notes.md file contains a routing table that instructs the agent to 'Run scripts/check-output.py' for output validation.
- Context: The provided script (scripts/check-output.py) is a benign utility that checks for content length and the presence of specific GTM-related keywords.
Audit Metadata