woostack-qa
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data from external URLs (web application content) and GitHub (PR/Issue descriptions and comments) to generate QA reports.
- Ingestion points: Data enters the agent context through the target web application at a user-supplied URL and through repository metadata/text from GitHub PRs, issues, or projects (documented in SKILL.md).
- Boundary markers: The skill includes explicit instructions to mitigate injection, stating that remote content, logs, and tool outputs must be treated as untrusted evidence, not instructions, and cannot be used to select tools or mutate the repository.
- Capability inventory: The agent is authorized to use browser capabilities, the GitHub CLI (gh), and write local markdown files to the .woostack/qa/ directory (documented in SKILL.md and references/report-template.md).
- Sanitization: The skill mandates a hard constraint to redact credentials, tokens, and personal data using placeholders like [REDACTED_TOKEN] before saving reports.
Audit Metadata