The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use a local script, scripts/resolve-source-root.mjs, to resolve the storage path for audit logs. This is a standard utility within the specified vendor environment.\n- [DATA_EXFILTRATION]: The instructions involve reading local environment variables (RAILLY_SKILLS_REPO) and checking specific home directory paths to determine where to store the TSV log. This data access is localized to the user's filesystem for the purpose of maintaining an audit trail and does not involve exfiltration to external networks.\n- [PROMPT_INJECTION]: The skill establishes an audit workflow where a secondary 'reviewer' model processes content generated by the primary agent. This ingestion of agent-generated text is a functional design feature to facilitate human or automated review of implementation choices.

trail-decisions