self-learning-agent-patterns
Pass
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is primarily educational and architectural documentation. It describes how agents can extract behavioral patterns (instincts) from user feedback and task outcomes.
- [DATA_EXPOSURE]: TypeScript examples access
process.env.SESSION_IDandprocess.env.AGENT_STRATEGY. This is used to tag internal metadata for learning signals within the agent's logic and does not involve exfiltration to external domains. - [COMMAND_EXECUTION]: No shell commands or system-level execution patterns are present in the skill instructions or code snippets.
- [INDIRECT_PROMPT_INJECTION]: While the skill describes systems that ingest user corrections, which is a theoretical attack surface for indirect injection in the final implemented system, the documentation itself promotes safety constraints, human-in-the-loop gates, and learning rate limits to mitigate such risks.
Audit Metadata