The Agent Skills Directory

[COMMAND_EXECUTION]: The skill implements a bash script (evaluate-session.sh) designed to run as a 'Stop' hook at the end of every Claude Code session. This script is used to automate administrative tasks such as counting messages and managing local file paths.
[PROMPT_INJECTION]: The skill features an indirect prompt injection surface inherent to its function of analyzing conversation transcripts.
Ingestion points: The evaluate-session.sh script identifies and accesses the session transcript path provided via stdin to calculate the message count.
Boundary markers: The instructions do not specify any delimiters or safety markers to isolate the session transcript content when it is evaluated by the agent for pattern extraction.
Capability inventory: The skill is designed to write new skill files to the ~/.claude/skills/learned/ directory based on the patterns identified during analysis.
Sanitization: No sanitization or filtering is performed on the transcript data before it is reviewed by the agent, meaning embedded instructions in the conversation history could theoretically influence the extraction process.

continuous-learning