liveview-streams
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill includes code patterns that process untrusted user input from browser events, creating a potential surface for indirect injection.
- Ingestion points: User-supplied parameters in handle_event('create', ...) and handle_event('save_edit', ...) in SKILL.md.
- Boundary markers: The HEEx template engine automatically escapes HTML content in <%= ... %> blocks.
- Capability inventory: The skill implements database operations (Repo.insert, Repo.update, Repo.delete) through a backend context module.
- Sanitization: Data is validated and cast through Ecto changesets before persistence.
- [DYNAMIC_EXECUTION]: The skill uses dynamic conversion of user-supplied strings to atoms for database sorting logic.
- Evidence: The implementation uses String.to_existing_atom(dir) and String.to_existing_atom(col) within handle_event('sort', ...) in SKILL.md. While safer than generating new atoms, this pattern allows users to influence database queries using any existing atom in the application runtime.
Audit Metadata