coven vs codeg
AI coding-agent runtimes and workspaces compared
coven and codeg both help developers run and coordinate AI coding agents, and both are written in Rust. coven focuses on project-scoped sessions managed by a local runtime, while codeg offers a broader workspace for agent conversations, code changes, and tasks across desktop, server, and mobile.
coven: Run Project-Scoped AI Coding-Agent Sessions
Coven is a Rust-based local runtime for project-scoped coding-agent sessions, with durable state and authority boundaries. It lets developers run Codex, Claude Code, GitHub Copilot CLI, and other harnesses on macOS, Linux, and Windows x64.
codeg: Coordinate AI Coding Agents with Agent Client Protocol
Codeg is a collaborative workspace for running and coordinating AI coding agents through the Agent Client Protocol. It brings agent conversations, sessions, code changes, and tasks into one experience across desktop, self-hosted server, Docker, and mobile.
| coven | codeg | |
|---|---|---|
| Language | Rust | Rust |
| License | MIT | Apache-2.0 |
| Stars | 73 | 4k |
| Forks | 26 | 518 |
| Last analyzed | Oct 28, 2026 | Oct 19, 2026 |
Key differences
- coven uses a local daemon to enforce project boundaries and persist session state; codeg organizes agent conversations, changes, and tasks in a shared workspace.
- coven focuses on Codex, Claude Code, and GitHub Copilot CLI; codeg connects 19 built-in agents and custom ACP agents.
- coven provides a local IPC API and CLI/TUI; codeg offers desktop, self-hosted server, Docker, mobile, and a CLI for scripts and CI.
- coven is MIT-licensed; codeg is Apache-2.0-licensed.
- coven is described as an early MVP with expected rough edges; codeg has a wider set of workspace features, with computer-use support identified as a preview.
- Both projects are Rust-based and have contributions concentrated in one person, a continuity consideration for teams assessing either project.
Choose coven if you…
- need a local runtime to manage project-scoped sessions across supported coding-agent harnesses.
- want session state and event logs persisted locally with a daemon enforcing project boundaries.
- prefer a CLI/TUI or a local IPC API for launching and observing agent sessions.
Choose codeg if you…
- want one workspace for conversations, code changes, project files, and agent tasks.
- need to delegate work across agents, including parallel tasks in isolated Git worktrees.
- want desktop, self-hosted, Docker, or mobile access, or a CLI for scripts and CI.
Compare the numbers side by side, or add more projects
This comparison is generated with AI from the OSRepos analyses of both projects. Always check each project's repository and documentation before choosing.