Monitors TrendR workflow runtime status and logs to detect stalls.
npx skills add https://github.com/gy-hou/trendr --skill trendr-watchdog
Please help me install this Agent Skill. Skill: trendr-watchdog Source: https://github.com/gy-hou/trendr/tree/main/skills/trendr-watchdog Command: npx skills add https://github.com/gy-hou/trendr --skill trendr-watchdog
运行时监督 TrendR 工作流，确保子任务完成后主会话继续跟进，检测进程卡死并自动注入断点续跑指令，以保持工作流连续性。
启动 trendr-watchdog，为 PROJECT 启动监控并在需要时自动注入续跑指令。
High-intent search queries and answers about installing and using this skill.
To prevent multi-phase agent orchestration from stalling overnight, you need an automated watchdog that monitors runtime logs for idle phases and injects resumption commands. This Skill applies phase-aware idle detection and cooldown-based resumes to ensure safe, automated recovery without progress loss.
Phase-aware idle detection works by continuously monitoring runtime status, progress, and log activity to identify unresponsive phases in agent workflows. When progress lags or hooks miss updates, the supervisor applies cooldown-based resumes from phase checkpoints to automatically recover the stalled orchestration.
Yes, you can automatically resume a stalled workflow from its last checkpoint. The watchdog detects when progress lags or hooks miss updates, then injects parameterized resumption commands into the owner session to restore the workflow from the exact phase checkpoint where it stalled.
The best way to monitor TrendR workflows for unresponsive states is using a dedicated supervisor that tracks runtime status and log activity. This Skill provides persistent overnight reporting and automatic stall detection, allowing you to review detailed watchdog reports for troubleshooting and post-mortem analysis.
No, you do not need any external dependencies to run automated workflow recovery. This Skill operates independently with zero dependencies, providing built-in phase-aware idle detection, cooldown-based resumes, and persistent reporting directly within your TrendR orchestration environment.
Multi-phase agent workflows lose progress after a stall when the owner session fails to follow up after subtasks complete. Without a watchdog injecting checkpoint resumption commands, the orchestration remains unresponsive and cannot automatically recover the lagging phases.
jason-c-devConfigures CloudWatch, CloudTrail, AWS Config, and Security Hub across accounts and regions.
AreteDriverDiagnose Gorgon workflow failures by inspecting checkpoint states, agent logs, and budget traces.
mkalkereCoordinate and monitor AI agents via HTTP with approval gates and dashboards.
robertzengcnRestore previously saved project work context across git branches.
