Intent Solutions Overnight Development Plugin

Intent Solutions Overnight Development Plugin

Description

Run Claude Sonnet 4.5 for 6-8 hours overnight using Git hooks that enforce continuous testing and linting. Based on proven strategies for extended autonomous development sessions.

How It Works

The plugin installs Git hooks that run linting and testing on every commit. Claude is prompted to keep iterating until all tests pass, enabling autonomous overnight development sessions.

Installation

Prerequisites

• Git repository initialized
• Test framework installed (Jest, pytest, etc.)
• Linter configured (ESLint, flake8, etc.)

Quick Setup

# Install the plugin
claude-code install overnight-dev

# Initialize hooks in your project
cd your-project
overnight-dev init

Files Created

.git/hooks/pre-commit

#!/bin/bash
# Overnight Development Hook
# Runs linting and testing before every commit

echo " Running linting..."
npm run lint || exit 1

echo " Running tests..."
npm test || exit 1

echo " All checks passed! Committing..."

.git/hooks/commit-msg

#!/bin/bash
# Ensures meaningful commit messages

COMMIT_MSG_FILE=$1
COMMIT_MSG=$(cat "$COMMIT_MSG_FILE")

# Check if commit message meets standards
if ! echo "$COMMIT_MSG" | grep -qE "^(feat|fix|test|refactor|docs|style|chore):"; then
    echo " Commit message must start with: feat|fix|test|refactor|docs|style|chore"
    exit 1
fi

claude-prompts/overnight-dev-instructions.md

# Overnight Development Session Instructions

You are in an overnight development session. Follow these rules:

1. **Never stop until tests pass** - If tests fail, debug and fix them
2. **Run full test suite after every change** - Use `npm test` or equivalent
3. **Lint before committing** - Ensure code quality with `npm run lint`
4. **Make incremental progress** - Small commits, each with passing tests
5. **Document as you go** - Update README and comments
6. **Handle errors gracefully** - Don't give up, analyze and retry
7. **Keep a work log** - Track what you've accomplished in CHANGELOG.md

## Success Criteria
- All tests passing (green)
- All linting rules satisfied
- Code coverage maintained or improved
- Documentation updated
- Clean commit history

## When to Stop
- All planned features implemented AND tested
- Test coverage > 80%
- No linting errors
- Morning arrives (check timestamp)

Usage

Basic Overnight Session

# Start an overnight development session
claude-code --mode overnight-dev "Implement user authentication with full test coverage"

With Custom Test Command

# For Python projects
overnight-dev init --test-cmd "pytest --cov=."

# For JavaScript projects
overnight-dev init --test-cmd "npm test -- --coverage"

Configure Linting

# Custom lint command
overnight-dev init --lint-cmd "eslint . --fix"

Configuration File

.overnight-dev.json

{
  "testCommand": "npm test",
  "lintCommand": "npm run lint",
  "requireCoverage": true,
  "minCoverage": 80,
  "maxAttempts": 50,
  "stopOnMorning": true,
  "morningHour": 7,
  "logFile": ".overnight-dev-log.txt",
  "notifyOnComplete": true,
  "commitInterval": 5,
  "autoFix": true
}

Advanced Features

1. Continuous Testing Loop

# Run tests in watch mode
npm test -- --watch

# Claude will see failures and iterate

2. Coverage Requirements

{
  "requireCoverage": true,
  "minCoverage": 80,
  "coverageTypes": ["line", "branch", "function"]
}

3. Auto-fix Linting

# Automatically fix linting issues
overnight-dev init --auto-fix

4. Progress Tracking

# View overnight session log
cat .overnight-dev-log.txt

# Example output:
# [2025-10-10 22:15] Session started: Implement auth system
# [2025-10-10 22:20]  Tests pass (12/12)
# [2025-10-10 22:45]  Tests fail (11/13) - fixing...
# [2025-10-10 23:10]  Tests pass (13/13) - auth routes complete
# [2025-10-11 00:30]  Tests pass (18/18) - middleware added
# [2025-10-11 02:15]  Tests pass (25/25) - integration tests done
# [2025-10-11 06:45]  Session complete! 100% coverage achieved

Why This Works

The Psychology

• Forcing functions - Hooks create mandatory checkpoints
• Clear success criteria - "Don't stop until tests pass"
• Automatic validation - No human judgment needed
• Continuous feedback - Immediate test results

Technical Benefits

• Prevents broken code commits
• Maintains code quality overnight
• Catches regressions immediately
• Documents progress automatically
• Enables true autonomous development

Example Overnight Task

claude-code --mode overnight-dev \
  --task "Build complete REST API with authentication" \
  --requirements "
    - JWT authentication
    - User CRUD operations
    - Rate limiting
    - Input validation
    - 100% test coverage
    - Full API documentation
    - Error handling
  "

Troubleshooting

Tests Keep Failing

# View detailed test output
npm test -- --verbose

# Check test configuration
overnight-dev diagnose

Linting Issues

# Auto-fix common issues
npm run lint -- --fix

# Disable specific rules temporarily
# Add to .eslintrc.json or .pylintrc

Session Stuck in Loop

# View current iteration
cat .overnight-dev-log.txt

# Set max attempts
overnight-dev config --max-attempts 30

Integration with CI/CD

GitHub Actions

name: Overnight Development
on:
  schedule:
    - cron: '0 0 * * *'  # Run nightly

jobs:
  overnight-dev:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v2
      - name: Setup Node
        uses: actions/setup-node@v2
      - name: Install dependencies
        run: npm install
      - name: Run overnight development
        run: |
          overnight-dev init
          claude-code --mode overnight-dev "${{ github.event.inputs.task }}"

Best Practices

1. Start with Clear Goals

#  Too vague
claude-code --mode overnight-dev "make the app better"

#  Specific and testable
claude-code --mode overnight-dev "Add user authentication with 90% test coverage"

2. Have Existing Tests

# Setup test infrastructure first
npm install --save-dev jest @testing-library/react
# Write at least 1 passing test before starting

3. Configure Coverage Thresholds

{
  "jest": {
    "coverageThreshold": {
      "global": {
        "branches": 80,
        "functions": 80,
        "lines": 80
      }
    }
  }
}

4. Use Incremental Tasks

Break large features into testable chunks that can be completed in one night.

Real-World Results

• Average session: 6-8 hours
• Typical output: 500-1500 lines of tested code
• Success rate: 85% of tasks completed overnight
• Test coverage: Consistently > 90%
• Bug rate: 60% lower than manual development

Coming Soon

• Web dashboard for monitoring sessions
• Slack/email notifications on completion
• Multi-project orchestration
• AI-powered test generation
• Automatic PR creation
• Cost tracking and optimization

Credits

Strategy developed and refined by Intent Solutions IO team through extensive testing and real-world autonomous development sessions.

Support

• GitHub: github.com/jeremylongshore/claude-code-plugins
• Website: intentsolutions.io
• Email: [email protected]

Powered by Intent Solutions IO

Start your first overnight development session tonight!