Debug software issues across context resets with structured session tracking.
npx skills add https://github.com/coco-research/coco --skill gsd-debug-coco-research
Please help me install this Agent Skill. Skill: gsd-debug Source: https://github.com/coco-research/coco/tree/main/systems/gsd/skills/gsd-debug Command: npx skills add https://github.com/coco-research/coco --skill gsd-debug-coco-research
This Skill helps diagnose and fix software issues that are hard to resolve in a single pass, especially when context resets, partial evidence, or long-running investigations make it difficult to keep track of what has already been tried.
Use the gsd-debug skill to investigate the reported issue, gather the necessary symptoms, and either diagnose the root cause or apply a fix.
High-intent search queries and answers about installing and using this skill.
To debug persistent failures across context resets, you need structured session tracking that preserves investigation state. This Skill uses checkpoint handling to save progress at decision points, allowing the debugging process to continue seamlessly even after context loss occurs.
The best way to isolate root cause analysis is by spawning a dedicated subagent for deeper investigation. This isolated agent debugging approach keeps the main conversation focused while gathering symptoms and performing complex troubleshooting separately to reach a confirmed root cause.
To gather symptoms effectively before investigating a bug, collect expected behavior, actual behavior, errors, timeline, and reproduction steps. This structured symptom gathering ensures the debugging agent has all necessary evidence to trace the issue and confirm the root cause before applying a fix.
Yes, you can resume a debugging investigation after a context loss using checkpoint handling. The Skill supports human verification and continuation prompts at decision points, allowing the isolated debugging agent to pick up the root cause analysis exactly where it left off.
Yes, structured debugging works for state management issues by using controlled Read, Bash, and Task tools to trace failures. It applies isolated subagent analysis to investigate recent changes and confirm the root cause of failing workflows before applying any fixes.
TheBushidoCollectiveDiagnose software bugs through evidence collection and hypothesis testing.
tomgunReproduce bugs with regression tests and apply minimal root-cause fixes.
SantiagoisperConnect Codex to a Chrome window for real-time monitoring and control.
SpecterShellManage persistent Python, Node, Bash, and Zsh REPL sessions in a daemon-backed environment.
