agentic-mcp-tools/owlex

Overview

Owlex is an MCP server that queries multiple CLI-based AI agents (Claude Code, Codex, Gemini, OpenCode) in parallel and runs deliberation rounds to synthesize their outputs. It is written in Python and designed for developers who want to compare or combine responses from different coding agents.

Best for

Best for
Developers who want to leverage multiple coding agents in parallel for comparison or consensus-driven results

Use cases

• Running the same prompt across multiple coding agents to compare answers
• Using deliberation rounds to refine or merge agent outputs for complex tasks
• Integrating multi-agent querying into existing MCP-compatible workflows

How to use

Install

uv tool install git+https://github.com/agentic-mcp-tools/owlex.git

Tools exposed

• security_audit
• code_review
• architecture_review
• devil_advocate
• start_codex_session
• resume_codex_session
• start_gemini_session
• resume_gemini_session
• start_opencode_session
• resume_opencode_session
• start_claudeor_session
• resume_claudeor_session
• start_aichat_session
• resume_aichat_session
• wait_for_task
• get_task_result
• list_tasks
• cancel_task
• COUNCIL_EXCLUDE_AGENTS
• COUNCIL_DEFAULT_TEAM

Tested with

Claude Code, Continue

Example client config

[object Object]

Notes

Owlex is an MCP server that queries multiple CLI-based AI agents (Claude Code, Codex, Gemini, OpenCode) in parallel and runs deliberation rounds to synthesize their outputs. It is written in Python and designed for developers who want to compare or combine responses from different coding agents.

121 stars on GitHub. Last updated 2026-03-15. Licensed MIT.

Use cases

• Running the same prompt across multiple coding agents to compare answers
• Using deliberation rounds to refine or merge agent outputs for complex tasks
• Integrating multi-agent querying into existing MCP-compatible workflows

Pros

• Supports four popular CLI agents out of the box
• Parallel execution reduces wait time for multi-agent queries
• Deliberation rounds can improve answer quality through cross-agent review

Cons

• Requires each CLI agent to be installed and configured separately
• Deliberation rounds add latency and may not always converge
• Limited to agents that expose a CLI interface

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.