Scoring was not performed

• Rename recipe__final_output to submit_final_output. Follow verb_noun pattern. Expand description to 80-120 chars explaining: this tool submits the final structured output to the user after validation against a predefined schema. Describe when to call it (when all reasoning is complete) and what happens on success/failure.
• Split platform__manage_schedule into 9 separate tools: list_scheduled_jobs, create_scheduled_job, run_scheduled_job_now, pause_scheduled_job, unpause_scheduled_job, delete_scheduled_job, kill_running_job, inspect_running_job, list_job_sessions, get_job_session_content. Each tool should do one thing with focused parameters and clear descriptions.
• For submit_final_output, document the output schema: what does the server return? Success response with confirmation ID? Error response with validation details? Include 2-3 example valid JSON structures in the description or linked docs.
• For each new scheduled_job tool, document required vs optional parameters. E.g., create_scheduled_job requires recipe_path and cron_expression; list_scheduled_jobs accepts optional limit parameter (default 50).
• Add error handling guidance. E.g., 'If job_id is invalid, the server returns a 404 error. Suggest the user call list_scheduled_jobs to see available job IDs.' This guides agent recovery.
• Document the cron expression format more clearly: '5-field (minute hour day month weekday) or 6-field (second minute hour day month weekday) formats. 5-field automatically converted to 6-field.' Provide 2-3 examples: '0 9 * * MON' (9 AM Mondays), '*/15 * * * *' (every 15 minutes).