wtk-qa-execute
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted data from human-generated bug reports and UAT results to guide QA activities.
- Ingestion points: Documentation and report paths including
docs/qa/README.md,docs/qa/reports/*.md,docs/qa/bugs/, and human UAT logs referenced inreferences/session-protocol.md. - Boundary markers: The skill requires the agent to verify all findings through an 'independent read path' and 'after a reload,' which helps ensure observations are not influenced by malicious data.
- Capability inventory: The agent uses browser adapters (Playwright MCP), CLI, and API interfaces, and has the authority to write reports, bug records, and evidence files to the local repository.
- Sanitization: The skill does not describe specific sanitization, filtering, or escaping logic for the external data it processes.
Audit Metadata