manage-cloud-build
Pass
Audited by Gen Agent Trust Hub on Nov 7, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from remote build logs, source code, and commit history. Evidence Chain: 1. Ingestion points: 'unity pipeline cloud-build builds log', 'git log', 'cm status', and project assets. 2. Boundary markers: Explicit instructions in SKILL.md to treat external content as data, not instructions. 3. Capability inventory: Bash CLI access and WebFetch. 4. Sanitization: Token masking and log-slicing via 'grep' and 'head'. This surface is well-mitigated.
- [DATA_EXFILTRATION]: The skill implements proactive data protection by masking JWT patterns in build logs and instructing the agent to use file-based password entry for signing credentials. This prevents sensitive credentials from being exposed in shell logs, process lists, or conversation history.
- [EXTERNAL_DOWNLOADS]: Documentation is retrieved via 'WebFetch' from official Unity domains (docs.unity.com and docs.unity3d.com). These are recognized as trusted vendor sources.
- [COMMAND_EXECUTION]: Shell commands are strictly limited to the 'unity' CLI and standard version control systems. State-changing operations (such as creating build targets) require explicit user approval, and interactive logins are forbidden to ensure the agent does not handle user sessions autonomously.
Audit Metadata