Nightshift automates codebase maintenance by leveraging AI (Claude/Codex) to identify and suggest fixes for issues like dead code, doc drift, and security vulnerabilities. It targets developers seeking proactive code quality improvements and efficient utilization of existing AI coding assistant budgets, offering a low-risk, PR-based workflow.

How It Works

The tool operates overnight, analyzing codebases using AI models to detect over 20 types of issues. It generates pull requests for all suggestions, allowing users to review and merge beneficial changes or discard unwanted ones, ensuring zero risk to primary branches. Nightshift is budget-aware, designed to consume only remaining daily AI allotments.

Quick Start & Requirements

Installation is available via Homebrew (brew install marcus/tap/nightshift), binary downloads, or go install. Prerequisites include the Claude Code or Codex CLIs, authenticated via subscriptions or API keys. A guided setup (nightshift setup) configures providers, projects, budgets, and daemon services. Official documentation for installation, quick start, and CLI reference is linked within the README.

Highlighted Details

• Budget-Aware Operation: Utilizes remaining daily AI budget, with configurable maximums (default 75%).
• Zero-Risk Workflow: All changes are proposed as branches or PRs, easily dismissible.
• Multi-Project Support: Configurable to monitor multiple repositories simultaneously.
• Comprehensive Issue Detection: Identifies dead code, doc drift, test gaps, security flaws, and more.
• AI Provider Flexibility: Supports both Claude and Codex AI models.

Maintenance & Community

The provided README does not detail specific contributors, community channels (e.g., Discord, Slack), sponsorships, or a public roadmap.

Licensing & Compatibility

Nightshift is released under the MIT License, which is generally permissive for commercial use and integration with closed-source projects.

Limitations & Caveats

The tool's functionality is dependent on the correct setup and authentication of external Claude Code and Codex CLIs. Certain provider configuration flags (dangerously_skip_permissions, dangerously_bypass_approvals_and_sandbox) suggest potential security considerations that require careful management. The effectiveness is also tied to the quality of the underlying AI models and user-defined configurations.