The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses a bash script (evaluate-session.sh) to perform lightweight session analysis at the end of a session. The script follows best practices, including safe path expansion and basic error handling, and limits its actions to local directory management and transcript reading.
[PROMPT_INJECTION]: The skill processes session transcripts to extract patterns, which technically exposes a surface for indirect prompt injection. However, this is the primary intended function of the skill, and the implementation does not bypass any of the agent's core safety features or instructions.

continuous-learning