charmbracelet/crush

Glamourous agentic coding for all 💘

View on GitHub ↗Jump to charts ↓Open shareable report →

Summary Information

Updated 1 hour ago
Added to GitGenius on September 2nd, 2026
Created on May 21st, 2025
Open Issues & Pull Requests: 697 (+1)
GitHub issues: Enabled
Number of forks: 2,232
Total Stargazers: 27,945 (+0)
Total Subscribers: 155 (+0)

Repository Insights (GitGenius)

Median issue/PR response: 16.6 hours
Mean response time: 10.4 days
90th percentile: 30.5 days
Tracked items: 1,023

How this project is maintained

Around half of the issues opened in the past year never receive a reply. 98% of open issues come from outside the core team, so the backlog reflects real-world use rather than internal planning. Work labelled "os: windows" is answered fastest, typically in about 6 hours, while "area: lsp" waits about 2 days. 56% of tracked open issues have had no activity in three months. Only 5% of issues opened in the past year have been closed.

Charts & Analytics

Fetching additional details & charts...

Issue Activity (beta)

Open issues: 415
New in 7 days: 13
Closed in 7 days: 4
Avg open age: 156 days
Stale 30+ days: 331
Stale 90+ days: 209

Recent activity

Opened in 7 days: 12
Closed in 7 days: 3
Comments in 7 days: 0
Events in 7 days: 0

Top labels

  • Migrated (142)
  • area: tools (123)
  • area: mcp (64)
  • area: session (64)
  • os: windows (49)
  • panic / crash (44)
  • provider: anthropic claude (34)
  • feedback (31)

Detailed Description

Crush is an agentic AI coding assistant that runs in the terminal and integrates with your choice of language models.

The tool addresses the need for AI-assisted coding that works within existing developer workflows and tooling. It connects language models to your local environment by integrating with Language Server Protocols for code context and Model Context Protocols for extending capabilities. This allows the LLM to understand your codebase and access tools via HTTP, stdio, and server-sent event transports. You can switch between different LLMs mid-session while preserving conversation context, and maintain multiple independent sessions per project.

Crush suits developers who want agentic AI assistance without leaving the terminal and who need flexibility in model selection. It works across macOS, Linux, Windows (both PowerShell and WSL), and several BSD variants. The tool supports a wide range of LLMs through OpenAI-compatible and Anthropic-compatible APIs, letting you bring your own model or choose from available options. The project is built on the Charm ecosystem, which powers thousands of applications in open source and production infrastructure.

Development activity shows consistent engagement with regular updates to the build system and active maintenance of core functionality. The project maintains broad platform support across multiple operating systems and continues to expand its capabilities through the MCP extension system.