better-interface
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to read and process untrusted workspace documentation (such as
CONTRIBUTING.md,AGENTS.md, andCLAUDE.md) and source code to understand project idioms. This behavior establishes an attack surface where malicious content in these files could attempt to influence the agent's review findings or behavior. - Ingestion points: Project documentation files (
CONTRIBUTING.md,CODING_STANDARDS.md,AGENTS.md,CLAUDE.md), Storybook documentation, and workspace source code. - Boundary markers: The skill includes defensive instructions such as 'A documented convention is no evidence the convention is good' and 'it's in the style guide does not retire a finding,' which instruct the agent not to treat untrusted data as authoritative override for quality standards.
- Capability inventory: Reading workspace files, inspecting the rendered state of the interface, and running project-provided test/check commands.
- Sanitization: The skill does not specify explicit methods for sanitizing or escaping the content read from these external files.
Audit Metadata