Install Zen MCP

Pick your client, copy the command, done.

claude mcp add -e "ZEN_API_KEY=${ZEN_API_KEY}" zen-mcp -- npx -y zen-mcp-server

Required:ZEN_API_KEY

Environment Variables

Set these before running Zen MCP.

VariableDescriptionRequired

ZEN_API_KEYAPI key for accessing the orchestration services.Yes

Once configured, Zen MCP gives your AI agent access to:

chatGeneral development chat and collaborative thinking.

messagemodel

thinkdeepExtended reasoning partner for complex problem solving.

query

challengeCritical challenge prompt to prevent superficial agreement.

statement

plannerInteractive step-by-step project planning.

goal

consensusMulti-model perspective gathering for analysis.

topicmodels

codereviewProfessional code review and audit.

code_path

precommitPre-commit validation for code changes.

files

debugExpert debugging assistant for complex problems.

error_log

After setup, try these prompts with your AI agent:

→Perform a codereview using gemini pro and o3, then use planner to generate a detailed plan for the identified fixes.

→Use the consensus tool to get perspectives from Claude and Gemini on the best architectural approach for this new feature.

→Debug the current codebase error using the debug tool and maintain the context for a follow-up refactor.

→Run a precommit validation on the modified files and ensure the output is reviewed by both Claude and O3.

Prerequisites & system requirements

An MCP-compatible client (Claude Code, Cursor, Windsurf, Claude Desktop, or Codex)
Node.js 18+ with npm/npx installed
ZEN_API_KEY — API key for accessing the orchestration services.

Conare · memory for coding agents

Save the docs, env vars, and workflow around Zen MCP in Conare so Claude Code, Codex, and Cursor remember it next time.