October 15, 2026

Use Chief From Your AI Assistant

Chief now has a remote MCP server. Connect it to Claude, Claude Code or Codex, and ask what's waiting on you, catch up on what shipped, or propose the next milestone from the assistant you already have open.

Why bring Chief into your assistant

Chief manages your agents across every project: it plans milestones, dispatches them to the coding agents on your own computer, reviews the results and collects the decisions only you can make. Most of the time, you check in on all of that from the Chief dashboard.

But a lot of your day happens somewhere else. You're in a Claude conversation sketching the next feature, in a Claude Code session in one repo, or in Codex working through a refactor. Now you can ask Chief from right there. The assistant talks to Chief over the Model Context Protocol, so the answer comes from your real projects, in the window you're already working in.

What you can ask

Ask in plain language. The assistant picks the right Chief tool for the question.

Catch up

  • "What's waiting on me?" The decisions, proposals and stalled milestones that need you, across the projects you shared.
  • "What changed since yesterday?" What happened since you last checked.
  • "Summarize this week on the billing project." The latest Chief summary for a project: daily, weekly or longer.
  • "What shipped today?" Milestones proposed, approved, completed or failed, and goal changes.

Look things up

  • "Show me the milestones about onboarding." Search milestones by status or text.
  • "What's the definition of done for the webhook milestone?" One milestone in full: definition of done, dependencies, status history and the latest agent result.
  • "What's on my to-do list in Chief?" Your manager tasks for a project.

Make changes

  • "Propose a milestone to add JSON output to every command." The proposal waits for your approval in Chief, like any other.
  • "Add a to-do to check the staging deploy." Add, dismiss or reopen your to-dos.
  • "Is my relay connected?" Whether the Chief relay on your computer is online, and help pairing it if it isn't.

Set it up

Every client uses the same server URL:

https://mcp.chiefisnext.com/mcp

Claude (web, desktop and mobile). In Claude's connector settings, choose Add custom connector, paste the server URL and connect. Sign in to Chief and pick your projects. Connectors you add on claude.ai also work in the Claude mobile apps.

Claude Code. Add the server, then run /mcp in Claude Code and choose Authenticate:

claude mcp add --transport http chief https://mcp.chiefisnext.com/mcp

Codex. Add the server and sign in. codex mcp login opens the Chief sign-in in your browser:

codex mcp add chief --url https://mcp.chiefisnext.com/mcp && codex mcp login chief

Any other assistant that supports remote MCP servers with OAuth sign-in can connect with the same URL. The Chief MCP server page has the current setup steps for each client.

You stay in control

  • You choose the projects. When you connect, you pick which projects the assistant can see. The rest stay invisible to it.
  • Proposals wait for you. A new connection can read and propose, but it can't approve or start work, change billing or settings, or show your code. Work still runs through the Chief relay on your own computer once you approve it.
  • Every change is labelled. Anything an assistant does shows up in Chief's activity feed with its name, like "Claude Code (MCP)".
  • Revoke at any time. See every connected assistant in Settings → Connected apps. Revoking takes effect on the assistant's next request.

Works on every plan

The MCP server works on every Chief plan, including Free. If you don't have an account yet, start free, connect a project, then add Chief to your assistant.

More articles

Multi-Agent Coding Orchestration Guide

Learn what multi-agent orchestration is, why developers are adopting it, and how to run 5+ AI coding agents in parallel without losing control. Free guide.

What Is AI-Native Development?

AI-native development isn't coding with AI assist — it's managing agents that build while you steer outcomes. Here's what that shift looks like in practice.

How to Review AI-Generated Code at Scale

AI agents write code faster than you can review it. Here are proven workflows for reviewing output from 3-5 agents without sacrificing quality or your sanity.

AI Coding Agent Costs in 2026

Claude Code $20–200/mo, Cursor $20/mo, Codex $200/mo, Aider BYOK: what running several AI coding agents really costs in 2026, and how to cut it.

From Single Agent to Multi-Agent

Already using one AI coding agent? Here's how to scale to multiple instances running in parallel — the concurrency patterns, worktree isolation, and task decomposition strategies that make it work.

Vibe Coding at Scale

Vibe coding with one agent is fun. Vibe coding with five is chaos — unless you have orchestration. Here's how to scale the vibe without losing the quality.

Why AI Agents Need an Orchestration Layer

Running AI coding agents in separate terminals is multiplexing, not orchestration. Here's why the difference matters — and why orchestration wins at scale.

Multi-Provider AI Orchestration

Claude Code, Codex, Cursor, Aider — each excels at different work. Here's how to combine providers into one coordinated workflow and why cross-provider review catches bugs single-provider setups miss.

Ready to manage your agents?

See everything Chief connects to on the [integrations page](/integrations).