advanced-debugging
Skill1 starsChanged 11 months agoMITView on GitHub
State-of-the-art debugging agent with hypothesis-driven analysis, automatic code instrumentation, git worktree isolation, and browser automation. Use when debugging errors, stack traces, unexpected behavior, performance issues, failed tests, race conditions, or hard-to-reproduce bugs.
- Deletes or force-pushes
In this repository4 other files it gives its agents.
- AGENTS.MDAGENTS.md
- oneplanSkill
- skill-generatorSkill
- tech-debt-reducerSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 3 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 109
Regression Test: Ensure no side effects
- Distinctive: line 152
When resolving a bug, ALWAYS provide:
- Distinctive: line 175
If after systematic analysis no solution is found:
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
🔍 Advanced Debugging System
A systematic, hypothesis-driven debugging approach with automatic instrumentation and isolated testing capabilities.
Quick Reference
| Task | Command |
| Instrument file | Run scripts/instrument.ts <file> |
| Analyze logs | Run scripts/analyze-logs.ts <logfile> |
| Create worktree | Run scripts/worktree.ts create <name> |
| Cleanup worktree | Run scripts/worktree.ts cleanup <name> |
| Generate hypothesis report | Follow "Hypothesis Generation" section |
Core Methodology
Phase 0: Hypothesis Generation (MANDATORY FIRST STEP)
NEVER attempt fixes before generating at least 3 hypotheses.
## 🔮 Bug Hypotheses
### Hypothesis 1: [Name] - Probability: [High/Medium/Low]
- **Suspected Cause**: [specific description]
- **Evidence Needed**: [what logs/tests to add]
- **Verification Method**: [how to confirm]
### Hypothesis 2: [Name] - Probability: [High/Medium/Low]
- **Suspected Cause**: [specific description]
- **Evidence Needed**: [what logs/tests to add]
- **Verification Method**: [how to confirm]
### Hypothesis 3: [Name] - Probability: [High/Medium/Low]
- **Suspected Cause**: [specific description]
- **Evidence Needed**: [what logs/tests to add]
- **Verification Method**: [how to confirm]Phase 1: Information Gathering
- Read the full error - Extract error type, message, stack trace
- Identify error category:
TYPE_ERROR- TypeScript/type issuesRUNTIME_ERROR- Execution failuresNETWORK_ERROR- API/fetch issuesDATABASE_ERROR- Prisma/SQL issuesBUILD_ERROR- Compilation/bundle issuesSTATE_ERROR- React state/hydration issues
Phase 2: Stack Trace Analysis
Error Analysis Template:
┌─────────────────────────────────────────────┐
│ ERROR: [error message] │
├─────────────────────────────────────────────┤
│ Type: [error type] │
│ First Project File: [file:line] │
│ Call Chain: fn1 → fn2 → fn3 │
│ Suspected Origin: [description] │
└─────────────────────────────────────────────┘Phase 3: Automatic Instrumentation
See INSTRUMENTATION.md for detailed patterns.
Quick instrumentation markers:
// 🔍 DEBUG markers for easy identification and cleanup
console.log('🔍 DEBUG [LOCATION]:', { timestamp: new Date().toISOString(), /* context */ });Cleanup pattern:
grep -rn "🔍 DEBUG" --include="*.ts" --include="*.tsx" src/Phase 4: Isolated Testing with Git Worktrees
For complex bugs, test hypotheses in parallel:
# Create isolated environment
git worktree add ../debug-hypothesis-1 -b debug/h1-[description]
# After finding solution
git worktree remove ../debug-hypothesis-1
git branch -D debug/h1-[description]See WORKTREES.md for multi-hypothesis workflows.
Phase 5: Fix-Verify-Document Cycle
- Minimal Fix: Change only what's necessary
- Verify: Reproduce original scenario
- Regression Test: Ensure no side effects
- Document: Comment non-obvious fixes
Error Pattern Quick Reference
See references/ERROR_PATTERNS.md for comprehensive patterns.
Common Next.js Patterns
| Error | Likely Cause | Quick Fix |
Hydration mismatch | Server/client content differs | Add suppressHydrationWarning or fix data |
Cannot read property of undefined | Missing null check | Add optional chaining ?. |
Module not found | Import path issue | Check tsconfig paths |
NEXT_REDIRECT | Redirect in wrong context | Use redirect() only in Server Components |
Common Prisma Patterns
| Error | Likely Cause | Quick Fix |
P2002 | Unique constraint violation | Check existing data |
P2025 | Record not found | Verify ID exists |
Connection refused | DB not running | Check DATABASE_URL |
Feedback Loop Integration
For UI bugs, use browser automation:
// Playwright verification loop
{
action: 'navigate', url: 'http://localhost:3000/[page]'
}
{
action: 'console_messages', errorsOnly: true
}
{
action: 'screenshot', fullPage: true
}Output Template
When resolving a bug, ALWAYS provide:
## 🐛 Debug Report
### 🔴 Problem Identified
[Clear description of the bug]
### 🔍 Root Cause
[Main cause with evidence]
### ✅ Solution Applied
[Fix with explanation]
### 🧪 Verification
[How to confirm fix works]
### 🛡️ Prevention
[How to avoid in future]Escalation Criteria
If after systematic analysis no solution is found:
- Summarize attempts made
- List remaining hypotheses
- Suggest next investigative steps
- Request additional context
Safety Constraints
- ❌ NO production data modification without confirmation
- ❌ NO destructive commands (
rm -rf,DROP DATABASE) without confirmation - ❌ NO ignoring warnings that indicate future problems
- ✅ ALWAYS ask confirmation for invasive fixes
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

