This project provides an opinionated set of extensions for Claude Code, designed to address common frustrations with AI-assisted development. It targets developers using Claude Code who want more control, predictability, and adherence to established patterns, ultimately aiming to improve the quality and maintainability of AI-generated code.

How It Works

The core of cc-sessions is its "brute authoritarian rule over the agentic loop" through a system of hooks and enforced protocols. These hooks intercept Claude's actions, requiring explicit user approval before code modifications (DAIC - Discussion, Analysis, Implementation, Confirmation), thus preventing unwanted refactors or broad changes. It also implements automatic context compaction to preserve session state and specialized sub-agents for tasks like context gathering and code review, ensuring Claude maintains awareness of the project's architecture and coding standards without constant re-explanation.

Quick Start & Requirements

• Installation: Install via npm/npx, pip/pipx, or a bash script. The installer guides through configuration, including user name and trigger phrases.
• Requirements: Claude Code, Python 3 + pip, Git.
• Setup Time: Approximately 5 minutes.
• Documentation: Customization details and reference guides are available within the sessions/knowledge/claude-code/ directory.

Highlighted Details

• Enforces a "discussion before implementation" loop via hooks that block write/edit tools.
• Implements automatic context compaction to preserve session state across restarts.
• Utilizes specialized agents (Context Gathering, Code Review, Logging) for improved context management and code quality checks.
• Enforces task-specific Git branches to prevent accidental modifications to the wrong codebase sections.
• Offers "Max Mode" for enhanced reasoning or "API Mode" for token savings, with features designed to reduce token consumption through context persistence.

Maintenance & Community

The project is a public good brought to you by GWUDCAP and Three AIrrows Capital. Customization is encouraged, with detailed guides on modifying hooks, agents, and configurations.

Licensing & Compatibility

The repository does not explicitly state a license in the provided README. Users should verify licensing for compatibility with their projects, especially for commercial use.

Limitations & Caveats

The README focuses heavily on the benefits and does not detail specific limitations or known issues. Users relying on specific AI model behaviors or advanced Claude Code features not covered by the hooks might encounter unexpected interactions. The lack of an explicit license could be a barrier for some organizations.