troubleshoot

Diagnose unexpected chat agent behavior by analyzing JSONL debug session logs.

Updated Sep 3, 2026
One-click install
npx skills add https://github.com/az0512124155azz-sys/CodeForge --skill troubleshoot-az0512124155azz-sys
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: troubleshoot
Source: https://github.com/az0512124155azz-sys/CodeForge/tree/main/extensions/copilot/assets/prompts/skills/troubleshoot
Command: npx skills add https://github.com/az0512124155azz-sys/CodeForge --skill troubleshoot-az0512124155azz-sys

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? When a chat agent behaves unexpectedly—slow requests, skipped tools, missing instructions, or failed tool calls—users have no visibility into what actually happened. This Skill investigates Copilot Chat debug logs to find evidence-based root causes instead of guessing. ## Core Features & Use Cases - Log-Based Root Cause Analysis: Parses session debug logs (main.jsonl, system prompts, tool definitions, subagent traces) to explain latency, errors, and unexpected model behavior. - Structured Investigation Workflow: Triages errors, slow events, discovery issues, and tool calls using grep/jq or PowerShell, then correlates events via span hierarchies. - Network Diagnostics & Fallbacks: Runs the Copilot diagnostics command for connectivity issues and consults the Copilot Issues wiki when no clear cause emerges. - Use Case: A user asks why their custom skill did not load. The Skill searches the session log, finds a name mismatch between the folder and SKILL.md frontmatter, and explains the fix. ## Quick Start Ask why your last chat request was slow or why a tool, skill, or agent did not load as expected.

Frequently Asked Questions about troubleshoot

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

FAQPage Schema
How do I find out why a Copilot Chat request was slow?▼

Search the session debug log for events with high duration values, such as filtering for dur greater than 5000 milliseconds. Check llm_request entries for time-to-first-token and token counts to identify whether latency came from the model or from tool execution.

Why did my custom skill or agent not load in chat?▼

Check the discovery events in the session debug log, which report how many instructions, skills, and agents were resolved and any skip reasons. A common cause is a mismatch between the folder name and the name field in the SKILL.md frontmatter.

How do I investigate a failed tool call in Copilot Chat?▼

Search the debug log for tool_call events with status error to see the tool arguments and error message. Correlate the event with its parent span to see which model response triggered the call.

Can I use workspace search tools to read Copilot debug logs?▼

No. Debug logs live outside the workspace in user storage, so workspace-scoped search tools cannot access them. Use terminal commands like grep, jq, Select-String, or Node.js one-liners instead.

What should I do when logs show network or authentication errors?▼

Run the github.copilot.debug.collectDiagnostics command, which returns a diagnostics report covering authentication status, network reachability, proxy settings, and certificates. Use that report to pinpoint the connectivity problem before concluding.

What if log analysis reveals no clear root cause?▼

Consult the Copilot Issues wiki on the microsoft/vscode GitHub repository for known issues matching the symptoms. If the wiki has no relevant guidance, report that the logs show no clear cause and suggest filing an issue.