software-delivery
Pass
Audited by Gen Agent Trust Hub on Aug 22, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill establishes structured guidelines for software engineering tasks such as planning, implementation, and debugging. It promotes safety best practices, such as requiring reproduction of failures before claiming fixes and implementing durable policy hooks for hazardous operations.- [PROMPT_INJECTION]: An indirect prompt injection attack surface exists due to the processing of potentially untrusted data.
- Ingestion points: The agent is instructed to read repository code, tests, instructions, and rendered browser content (SKILL.md, browser-and-e2e.md).
- Boundary markers: There are no explicit instructions to use delimiters or ignore instructions embedded within the ingested repository data or browser pages.
- Capability inventory: The skill directs the agent to implement code changes, execute testing suites, and run diagnostic commands across the workspace.
- Sanitization: The guidance includes a safety mitigation to protect secrets and personal data when capturing failure evidence like traces and screenshots.
Audit Metadata