The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to apply executable permissions using chmod +x and perform validation by executing the generated shell scripts.
[DYNAMIC_EXECUTION]: The skill dynamically generates Bash scripts at runtime within the ~/.claude/scripts/ directory and subsequently executes them for verification.
[PERSISTENCE]: The skill creates persistent executable files in a hidden user directory and references cron-bot-tasks.sh for use with the system cron scheduler, enabling automated execution across sessions.
[INDIRECT_PROMPT_INJECTION]: The generated scripts process external and potentially untrusted data, which could be used to influence script behavior if the data contains malicious payloads.
Ingestion points: The scripts ingest data from CLAUDE.md files, repository names, and bot notification/comment context.
Boundary markers: There are no defined boundary markers or delimiters for the external data being parsed.
Capability inventory: The skill possesses file-write access and the ability to execute shell scripts on the local system.
Sanitization: The skill lacks explicit sanitization or validation routines for the data processed by the library functions or generated scripts.

setup-scripts