Agent Launch – One CLI for Codex, Claude Code, Cursor, Gemini, OpenCode
Unified CLI for coding agents, but shell aliases or a simple script could do the same.
Use Gemini from Claude Code to review code or delegate tasks.
Session-oriented ACP runtime differs from codex-plugin-cc's app-server approach.
Developers using both Claude Code and Gemini CLI
codex-plugin-cc · Claude Code plugins
https://github.com/abiswas97/gemini-plugin-cc
The repo is derived from openai/codex-plugin-cc, but the runtime layer is different: this plugin talks directly to Gemini CLI in ACP mode instead of Codex app-server.
In practice, the Gemini side here is much more session-oriented: - spawn `gemini --acp` (or `--experimental-acp` for older CLI versions) - initialize - create or load a session - set mode / model - send prompts - stream `session/update` - handle file-system / permission callbacks - cancel when needed
That difference shows up in the plugin design too. Codex app-server has richer built-in concepts around review / turn lifecycles. In this repo, review is implemented in the plugin layer on top of the generic ACP task flow.
Current commands are: `/gemini:review`, `/gemini:adversarial-review`, `/gemini:task`, `/gemini:rescue`, plus status / result / cancel.
Built with Claude Code + Codex, which felt appropriately recursive.
Unified CLI for coding agents, but shell aliases or a simple script could do the same.
Yet another LLM orchestrator when MCP already enables this natively.
Docker isolation for AI agents solves the death-by-tapping-y problem safely.
Surfaces model disagreements instead of averaging them away — that's the real value.
Adversarial plan review via hook injection—clever constraint on existing tools.
Meta-agents can control Circus Chief itself, a rare self-referential twist.