uniform-forms
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines a system for ingesting untrusted user data via form fields. It includes mandatory guidance on mitigating risks associated with this attack surface.
- Ingestion points: Defined in
references/submission.mdas a JSON payload containing user-supplied labels and values. - Boundary markers: The use of structured JSON for the submission payload provides clear data boundaries.
- Capability inventory: The skill instructs the agent/developer to perform a POST request to a submission endpoint (
SKILL.md). - Sanitization: The
references/submission.mdfile explicitly states that users must 'Validate and sanitise each item of input data before storing or sending it out.' - [SAFE]: The skill contains no malicious code, obfuscation, or unauthorized commands. It follows industry-standard patterns for form modeling and submission handling, including warnings against unnecessary sensitive data collection (e.g., password fields).
Audit Metadata