retro
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted session logs and repository files to identify improvements. While this represents a data ingestion surface, the skill has no destructive capabilities (it only presents candidates to the user), uses no automated write tools, and is restricted by 'disable-model-invocation: true' in its metadata, which prevents the agent from making autonomous tool calls based on the ingested content.
- [COMMAND_EXECUTION]: The skill instructions advise searching through session logs on the local machine. This is a standard analytical task for a retrospective tool and does not involve elevated privileges or dangerous shell commands.
- [DATA_EXPOSURE]: The skill reads repository configuration files such as 'package.json', CI workflows, and internal steering files ('AGENTS.md', 'CODING_STANDARDS.md'). This access is limited to the repository context and is necessary for its stated purpose of environment improvement.
Audit Metadata