diagnose

Diagnose

Systematically investigate a problem using evidence-based reasoning. This skill follows a structured investigation methodology to find root causes and propose solutions.

Arguments

• {problem} -- Description of the problem to investigate (required)
• --quick -- Fast diagnosis, skip deep analysis
• --logs -- Include log analysis in investigation

Process

Phase 1: Problem Statement

Clarify the problem:

• What is the expected behavior?
• What is the actual behavior?
• When did it start? (commit, deployment, time)
• Is it reproducible? How?
• What's the impact? (severity, affected users)

Phase 2: Evidence Collection

Use an investigator approach to gather evidence systematically.

Source 1: Error Messages & Logs

# Recent git history
git log --oneline -20

# Search for related errors in codebase
grep -r "ERROR_CODE" src/

# If --logs flag, search log patterns
grep -rn "{error pattern}" logs/

Source 2: Code Analysis

# Find related code
grep -rn "{symptom keyword}" src/

# Check recent changes to affected area
git log -p --since="1 week ago" -- {affected paths}

# Find usages and dependencies
grep -rn "{function/class name}" src/

Source 3: Configuration

# Check appsettings
cat src/**/appsettings*.json | grep -i "{related config}"

# Check environment variables
grep -rn "GetEnvironmentVariable\|GetValue<" src/ | grep -i "{related}"

Source 4: Tests

# Find related tests
grep -rn "{feature}" tests/

# Check if tests are passing
dotnet test --filter "{test pattern}" --no-build

Source 5: External Context

• Search for similar issues in GitHub issues
• Search for related error messages online
• Check documentation for expected behavior
• Use context7 to look up library docs when the error involves a specific library -- this often reveals known issues or correct usage patterns faster than web search

Phase 3: Hypothesis Generation

Based on evidence, generate hypotheses ranked by likelihood: