pipeline-window-control

Monitor and control pipeline sub-window processes via PID tracking and DB status checks.

10|3|Updated Mar 7, 2026
One-click install
npx skills add https://github.com/Cynthia1070711/PHYCOOL_Tools --skill pipeline-window-control-cynthia1070711
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: pipeline-window-control
Source: https://github.com/Cynthia1070711/PHYCOOL_Tools/tree/main/config-templates/claude/skills/pipeline-window-control
Command: npx skills add https://github.com/Cynthia1070711/PHYCOOL_Tools --skill pipeline-window-control-cynthia1070711

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires better-sqlite3, and includes scripts (resource) components.

What problem does it solve? When running multi-story development pipelines, sub-windows can become zombies (task finished in the database but the process still runs) or hang indefinitely, and the central controller lacks visibility into which windows are active, stuck, or complete. ## Core Features & Use Cases - Active Window Status: List all tracked pipeline sub-windows with PID liveness, database story status, phase target, and elapsed time via --status. - Zombie Detection & Cleanup: Detect windows whose DB status reached the phase target but whose process still runs, then gracefully close them with --check-zombie. - Manual & Emergency Control: Force-close a specific story's window with --kill {story_id} or all windows with --kill-all, and read completion notifications written by the Stop Hook via --notify. - Use Case: After launching a batch of pipeline stories, the controller runs --status to verify progress, then --check-zombie between batches to clean up finished-but-open windows before dispatching the next batch. ## Quick Start Ask the assistant to run the pipeline-control script with the --status flag to show all currently active pipeline sub-windows and their database status.

Frequently Asked Questions about pipeline-window-control

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I check the status of running pipeline sub-windows?

Run node pipeline-control.js --status to print a table of all tracked windows showing story ID, phase, PID, process liveness, database status, phase target, and elapsed minutes. It also warns if any zombie windows are detected.

How do I detect and kill zombie processes after a pipeline batch?

Run node pipeline-control.js --check-zombie. A zombie is a window whose database story status already reached the phase target but whose PID is still running; the script gracefully closes child processes first, then force-kills the process tree if needed.

What is the difference between pipeline-control.js and close-worker.ps1?

They are two parallel systems with separate ledgers. pipeline-control.js tracks PIDs in logs/pipeline-active.json, while close-worker.ps1 manages worker_runs database rows with five preconditions and CAS final-state writes. Windows registered in worker_runs must be closed via close-worker.ps1, never with --kill.

Does pipeline-control.js work on Linux or macOS?

No. The script relies on Windows-specific tooling: PowerShell Get-Process and Get-CimInstance for liveness checks and child-process termination, plus taskkill for force-killing process trees. It also requires better-sqlite3 to read the story database.

Why does --kill not close a window registered in the worker_runs table?

pipeline-control.js only reads the logs/pipeline-active.json tracker and has no knowledge of the worker_runs database table. The two tracking ledgers are not synchronized, so killing by PID leaves the worker_runs lifecycle state inconsistent.