Why This Matters

Bug triage is a daily reality. An assistant that takes a stack trace, diagnoses the root cause, and proposes a fix saves hours per week per developer.

Concept Explained

A CLI tool or Chat workflow: paste a stack trace. Copilot diagnoses root cause, identifies affected files, proposes a fix. Output as JSON for pipeline integration.

How It Works

Build a CLI script that takes a stack trace (stdin or file). Calls Copilot with a schema: {severity, category, likely_cause, suggested_fix, affected_files, confidence}. Outputs JSON.

Step-by-Step Tutorial

1. Define schema

2. Build CLI script

3. Validate output

4. Test on real traces

5. Integrate with issue tracker

Real-World Example

A team built a bug-triage assistant. Run on every error log in production. Filed issues automatically with severity, category, and suggested fix. Saved ~30 minutes per incident. Cut mean time to resolution by 40%.

Example Prompts / Commands / Code

#!/usr/bin/env python3
# triage.py
import subprocess
import json
import sys

schema = '''
Output as JSON matching this TypeScript type:
type BugAnalysis = {
  severity: 'high' | 'medium' | 'low';
  category: 'runtime' | 'logic' | 'security' | 'performance' | 'other';
  likely_cause: string;
  suggested_fix: string;
  affected_files: string[];
  confidence: number;
};
Output ONLY the JSON. No prose.
'''

stack_trace = sys.stdin.read()
prompt = f'Analyze this stack trace.\n\n{stack_trace}\n\n{schema}'

result = subprocess.run(['copilot', '--no-interactive', prompt],
                       capture_output=True, text=True)

try:
    analysis = json.loads(result.stdout)
    print(json.dumps(analysis, indent=2))
except:
    print('Failed to parse Copilot output as JSON', file=sys.stderr)
    sys.exit(1)

$ cat stack-trace.log | python3 triage.py
{
  "severity": "medium",
  "category": "runtime",
  "likely_cause": "Null reference in user lookup when id is empty",
  "suggested_fix": "Add null check before user lookup in auth/service.ts:42",
  "affected_files": ["src/auth/service.ts"],
  "confidence": 0.85
}

Common Mistakes

• No schema — output is unstructured prose.
• No validation — silent JSON parse failures.
• Trusting diagnoses without verification.
• Not testing on real traces.

Best Practices

• Define a clear JSON schema.
• Validate output programmatically.
• Test on 10+ historical traces.
• Verify diagnoses against actual code.
• Integrate with issue tracker for automation.

Troubleshooting

Practical Exercise

Professional Challenge

Frequently Asked Questions

Further Reading

• Copilot CLI

Official References

• Copilot CLI

Related lessons: BE-29, IN-04