agent-orchestration-improve-agent
Pass
Audited by Gen Agent Trust Hub on Apr 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill defines a logical workflow for agent optimization without incorporating malicious instructions or unauthorized data access.
- [COMMAND_EXECUTION]: The use of the analyze-agent-performance tool with $ARGUMENTS is a standard way to parameterize tool calls within the supported agent environments.
- [DATA_EXFILTRATION]: While the skill involves analyzing performance data and user feedback, these actions are central to its stated purpose of agent improvement and do not involve exfiltrating sensitive information to external parties.
- [PROMPT_INJECTION]: The skill involves analyzing user feedback, creating a surface for indirect prompt injection. The instructions focus on pattern analysis and classification, which inherently limits the risk of executing untrusted instructions.
Audit Metadata