apple-container-coding-agent
Pass
Audited by Gen Agent Trust Hub on Aug 26, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes
subprocess.runto call thegitCLI for snapshot management, patch inspection, and diffing. It also uses this mechanism to launch the AI agent within the isolated container environment. Specifically,scripts/prepare-snapshot.pyandscripts/inspect-patch.pyuse these calls for repository maintenance, whileassets/worker-entrypoint.pymanages the agent process execution.- [EXTERNAL_DOWNLOADS]: Theassets/Containerfilefetches AI agent packages from the NPM registry during the build phase. The documentation advises specifying exact versions to maintain environment integrity.- [DATA_EXFILTRATION]: The skill includes defensive features to prevent exfiltration, such as thescripts/prepare-snapshot.pyutility which scans for and blocks sensitive files (e.g., keys, credentials, and environment files) from being included in the agent's work environment.
Audit Metadata