code-debugging

Included with Lifetime

$97 forever

Debug experiment code with structured error analysis. Categorize errors, apply targeted fixes with retry logic, and use reflection to prevent recurring issues. Use when experiment code fails or produces incorrect results.

Code Review

What this skill does


# Code Debugging

Systematically debug experiment code with structured error categorization and fix strategies.

## Input

- `$0` — Error message, stderr output, or code file with issues
- `$1` — Optional: the code that produced the error

## References

- Debug patterns and state machine: `~/.claude/skills/code-debugging/references/debug-patterns.md`

## Workflow

### Step 1: Categorize the Error

| Category | Examples | Severity |
|----------|----------|----------|
| SyntaxError | Invalid syntax, indentation | Low |
| ImportError | Missing module, wrong name | Low |
| RuntimeError | Division by zero, shape mismatch | Medium |
| TimeoutError | Infinite loop, too slow | Medium |
| OutputError | Missing files, wrong format | Medium |
| LogicError | Wrong results, 0% accuracy | High |

### Step 2: Analyze Root Cause
1. Read the error traceback (last 1500 chars if truncated)
2. Identify the exact line and variable causing the error
3. Check for common patterns:
   - Device mismatch (CPU vs GPU tensors)
   - Shape mismatch in matrix operations
   - Missing data normalization
   - Off-by-one errors in indexing
   - Incorrect loss function for task type

### Step 3: Apply Fix Strategy

**For syntax/import errors**: Direct fix, single attempt
**For runtime errors**: Fix and rerun, up to 4 retries
**For logic errors**: Reflect on approach, consider alternative methods
**For timeout**: Reduce dataset size, optimize bottleneck, add early stopping

### Step 4: Reflect and Prevent
After fixing:
1. Explain why the error occurred
2. Identify which lines caused it
3. Describe the fix line-by-line
4. Note patterns to avoid in future code

## Fix Strategy State Machine

```
Stage 0 (first attempt) → repost code as fresh
Stage 1 (second attempt) → repost or leave depending on severity
Stage 2 (third attempt) → regenerate from scratch if still failing
```

## Rules

- Prefer minimal targeted edits over full rewrites
- Maximum 4-5 fix attempts before changing approach
- Always truncate long error outputs to last 1500 characters
- After fixing, verify the fix doesn't introduce new errors
- Keep error history to avoid repeating the same mistakes
- If 0% accuracy: check accuracy calculation first, then check data pipeline

## Related Skills
- Upstream: [experiment-code](../experiment-code/)
- See also: [paper-to-code](../paper-to-code/), [data-analysis](../data-analysis/)

Files: 2

Size: 6.8 KB

Complexity: 25/100

Category: Code Review

Source: https://github.com/lingzhi227/agent-research-skills/tree/main/skills/code-debugging

Related in Code Review

gstack

Included

Fast headless browser for QA testing and site dogfooding. Navigate pages, interact with elements, verify state, diff before/after, take annotated screenshots, test responsive layouts, forms, uploads, dialogs, and capture bug evidence. Use when asked to open or test a site, verify a deployment, dogfood a user flow, or file a bug with screenshots. (gstack)

Code Reviewscriptsfeatured

startup-due-diligence

Included

Legal due diligence review for seed-stage and Series A startups (US, Delaware C-Corp focus). Supports both investor and founder perspectives. Capabilities include: (1) Interactive document review and issue spotting; (2) Document request list generation; (3) Cap table and SAFE/convertible note analysis; (4) Red flag identification with severity ratings; (5) Diligence report generation. TRIGGERS: due diligence, DD, startup investment, cap table review, Series A, seed round, investor diligence, legal review startup, SAFE analysis, convertible note, 409A, founder vesting.

Code Reviewscripts

interview-master

Included

This skill should be used when the user asks to "generate interview questions", "prepare for interview", "optimize resume", "conduct mock interview", "analyze git commits for resume", "generate resume from code", "review my resume", or mentions interview preparation, career assistance, or extracting project experience from git history. Provides comprehensive interview and career development guidance for both job seekers and interviewers.

Code Reviewscripts

fix-issue

Included

Fixes GitHub issues using parallel analysis agents for root cause investigation, code exploration, and regression detection. Reads issue context from gh CLI, searches codebase and memory for related patterns, generates a fix with tests, and links the resolution back to the issue via PR. Includes prevention analysis to avoid recurrence. Use when debugging errors, resolving regressions, fixing bugs, or triaging issues.

Code Reviewscripts

sf-apex

Included

Generates and reviews Salesforce Apex code with 150-point scoring. TRIGGER when: user writes, reviews, or fixes Apex classes, triggers, test classes, batch/queueable/schedulable jobs, or touches .cls/.trigger files. DO NOT TRIGGER when: LWC JavaScript (use sf-lwc), Flow XML (use sf-flow), SOQL-only queries (use sf-soql), or non-Salesforce code.

Code Reviewscripts

swift-development

Included

Comprehensive Swift development for building, testing, and deploying iOS/macOS applications. Use when Claude needs to: (1) Build Swift packages or Xcode projects from command line, (2) Run tests with XCTest or Swift Testing framework, (3) Manage iOS simulators with simctl, (4) Handle code signing, provisioning profiles, and app distribution, (5) Format or lint Swift code with SwiftFormat/SwiftLint, (6) Work with Swift Package Manager (SPM), (7) Implement Swift 6 concurrency patterns (async/await, actors, Sendable), (8) Create SwiftUI views with MVVM architecture, (9) Set up Core Data or SwiftData persistence, or any other Swift/iOS/macOS development tasks.

Code Reviewscripts