Agent shell is what I always wanted. I have been using many of the different Claude code integrations packages and they are really good. But there is always some friction because I need to run it in a terminal emulator. With agent shell it feels so much more integrated and natural.
I am really excited for these improvements, especially reading the env from a file.
I wish that agent-shell-sidebar had some screenshots though so I could see what it actually does.
agent-shell: A single native Emacs experience to interact with different AI agents powered by ACP (Agent Client Protocol) https://agentclientprotocol.com
So far, agent-shell can interact with Claude Code, Gemini CLI, Codex, and Goose, but can technically work with any ACP-powered agent.
I think the difference is ECA is a coding agent with a LSP-like protocol for various frontend and editors, which itself supports many models.
Where as agent protocol if I understand lets you use many agents like Gemini CLI, Claude Code, well assuming they support the protocol, using various frontend?
Though I guess other coding agents could also adopt the ECA protocol maybe.
I am really excited for these improvements, especially reading the env from a file.
I wish that agent-shell-sidebar had some screenshots though so I could see what it actually does.
So far, agent-shell can interact with Claude Code, Gemini CLI, Codex, and Goose, but can technically work with any ACP-powered agent.
ps. agent-shell needs more sponsors to become sustainable https://github.com/sponsors/xenodium
I think the difference is ECA is a coding agent with a LSP-like protocol for various frontend and editors, which itself supports many models.
Where as agent protocol if I understand lets you use many agents like Gemini CLI, Claude Code, well assuming they support the protocol, using various frontend?
Though I guess other coding agents could also adopt the ECA protocol maybe.
Come on, you unknown hero!
(and thanks to the Zed team and Google for building the spec)