- Deliberation
Deliberation
What it is
Deliberation connects GPT (via the Codex CLI), Gemini (via the Antigravity CLI agy), Grok (xAI), and any OpenAI-compatible model through OpenRouter into one MCP server.
You ask once; it routes the question to whichever providers you have configured and synthesizes their answers instead of dumping raw output. Use any subset - it detects what is set up and routes accordingly.
Seven experts cover the work, each with a focus: Architect, Plan Reviewer, Scope Analyst, Code Reviewer, Security Analyst, Researcher, Debugger. GPT and Gemini can also implement changes; Grok and OpenRouter advise only.
Install
Any MCP host (stdio):
{ "mcpServers": { "deliberation": { "command": "npx", "args": ["-y", "@antonbabenko/deliberation-mcp"] } } }
In Claude Code, install the plugin instead:
/plugin marketplace add antonbabenko/agent-plugins
/plugin install deliberation@antonbabenko
/deliberation:setup
Tools
ask-all- send one question to every configured provider in parallel, then compare the answersconsensus- providers review while the host commits a blind verdict, then adjudicates; converges only when models agreeask-gpt/ask-gemini/ask-grok/ask-openrouter- a single-provider second opinion- the seven experts, callable directly
Requirements
You need at least one provider:
- GPT: the Codex CLI (
npm i -g @openai/codex, thencodex login) - Gemini: the Antigravity CLI (
agy), logged in - Grok:
XAI_API_KEY(get a key at https://console.x.ai) - OpenRouter:
OPENROUTER_API_KEYplus models declared in~/.claude/deliberation/config.json; works with any OpenAI-compatible endpoint (Ollama, vLLM, LM Studio)
Node 18+. The server is bundled and ships with zero runtime dependencies.
Server Config
{
"mcpServers": {
"deliberation": {
"command": "npx",
"args": [
"-y",
"@antonbabenko/deliberation-mcp"
],
"env": {
"XAI_API_KEY": "<optional, for Grok>",
"OPENROUTER_API_KEY": "<optional, for OpenRouter>"
}
}
}
}Recommend Servers
View All高德地图官方 MCP Server
Playwright MCP server
Write notes to Flomo
summarize chat message