exploratory-testing
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the application's UI, API, and database, which creates a potential surface for indirect prompt injection. * Ingestion points: The Recon phase retrieves data from databases using [DB_TOOL], API endpoints via [API_TOOL], and the user interface using playwright-cli. * Boundary markers: The skill implements a 'Refute' phase and references the adversarial-verification.md doctrine to validate findings, providing a logical check against malicious data triggers. * Capability inventory: The agent possesses significant capabilities, including database access, authenticated API calls, browser automation, and integration with the issue tracker (acli). * Sanitization: The provided instructions do not detail specific sanitization or escaping procedures for retrieved application data.
- [COMMAND_EXECUTION]: The skill executes various local project-specific scripts via 'bun run', including context:coverage, api:login, context:map, and jira:sync-issues, to manage testing context and credentials.
Audit Metadata