reflect
Pass
Audited by Gen Agent Trust Hub on May 16, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill facilitates a reflection process that ingests untrusted data from the conversation history to modify the agent's knowledge base and instruction set.
- Ingestion points: The 'Process' section in SKILL.md explicitly instructs the agent to 'Scan the conversation' for knowledge, mistakes, and corrections, which are provided by the user.
- Boundary markers: The skill lacks explicit boundary markers or instructions to differentiate between helpful user feedback and adversarial prompt injection attempts embedded in the dialogue.
- Capability inventory: The 'Routing' section grants the agent the ability to write to '.agents/skills/' and 'brain/' directories, allowing it to modify its own future system instructions and behavioral logic.
- Sanitization: No sanitization, escaping, or validation steps are defined to ensure that the content extracted from the conversation is safe before being persisted into the agent's configuration files.
Audit Metadata