The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest conversation history to update long-term knowledge files. This introduces a vulnerability surface where malicious instructions planted in the conversation could be persisted into the agent's 'brain', potentially influencing future behavior through gradual poisoning.
[DATA_EXPOSURE]: The skill reads from and writes to the brain/ directory. This is consistent with its stated purpose of knowledge management and does not target sensitive system files or credentials.
[SAFE]: No obfuscation, remote code execution, or unauthorized command execution patterns were identified in the provided instructions.

reflect