crypto-auto-progression_skill

Overview

This skill enables and maintains real-execution-driven automatic progression for the crypto-hedge-backtest project. It uses cron-driven tasks that perform real work on each trigger, report only when stage成果 exist, and provide health checks and daily summaries. The automation is designed to be reliable, observable, and repeatable.

How this skill works

The skill sets up a primary 5-minute cron job that performs at least one real execution per run (run scripts, modify code, or produce files) and includes stage-result reporting. It adds a 30-minute guardian job to verify recent outputs and alert on blocking issues, plus an optional daily summary cron. Creation follows a strict order and includes immediate forced run and verification steps to ensure correctness.

When to use it

• Enable automated progression every N minutes that performs real actions rather than reminders.
• Require periodic health checks to detect stalled pipelines or missing outputs.
• Deliver daily summaries of completed items, risks, and next steps to stakeholders.
• Diagnose recurring cron failures or parameter errors in cron job creation.
• Convert an ad‑hoc cron workflow into a reusable, auditable automation pattern.

Best practices

• Always create the 5-minute primary job first, then force a run and confirm status before adding other jobs.
• Ensure each cron payload emphasizes real execution, stage-result reporting, and suppresses redundant no-result notifications.
• Implement network retries with exponential backoff for external data pulls (e.g., Binance) before failing the scan path.
• Verify cron job fields: name, schedule, sessionTarget, and payload must be present and non-empty.
• Use the two allowed status phrases for reports to keep operator expectations consistent.

Example use cases

• Automate backtest progression so every trigger runs a script that produces a new results file or checkpoint.
• Periodic health guard that checks for new commits/files in the last 30 minutes and alerts with clear blocking reasons.
• Daily 21:30 Asia/Manila summary listing completed work, key metrics, risks, and tomorrow’s plan.
• Troubleshoot and fix recurring cron.add errors by re-sending a complete job object instead of retrying a malformed request.
• Harden data fetches from exchanges by adding retry and exponential backoff to prevent transient SSL/network failures.