The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes session transcripts containing potentially untrusted user input. While the skill primarily focuses on pattern extraction, instructions embedded within a conversation could theoretically influence the agent during the evaluation phase. However, given the local processing and the specific intended purpose, this risk is considered low and inherent to the functionality.
[COMMAND_EXECUTION]: The skill utilizes a local shell script (evaluate-session.sh) as a 'Stop' hook. The script performs basic operations such as checking message counts and directory creation, following best practices like quoting variables and using standard utilities (jq, grep).
[SAFE]: No indicators of credential theft, data exfiltration, or malicious obfuscation were found. The skill's behavior is consistent with its stated purpose of improving agent performance through session analysis.

continuous-learning