Gemini + Steady

Gemini integration for Steady

Gemini CLI puts a Gemini-powered assistant inside the terminal you already live in. Pointing it at Steady’s MCP server makes your check-ins, Goal Stories, team activity, and digest first-class participants in that session — no separate tab, no copy-pasting context, no scraping the API yourself.

It’s a particularly natural fit for engineering workflows. The data you typically want — what your teammates reported this week, what’s outstanding on a goal, who’s out tomorrow — lands directly in the same surface you’re using to read code, run jobs, and shape commits.

Where this fits

Reach into Steady from any prompt. Daily Digest, Smart Check-ins, Goal Stories with their full update history, Echoes, activity feed, insights, absences — Gemini CLI can pull any of it on demand. Ask in plain English, refine with follow-ups, and stay in the conversation.

Send work back to Steady. Gemini can post check-ins, open Goal Stories, and add new updates to existing ones. Useful at the end of a working session: narrate what shipped and what’s blocked, then have Gemini file the check-in for you.

Wire it into automations. Because Gemini CLI is scriptable, the same MCP connection drives unattended jobs. Generate a weekly engineering report from Goal Story updates. Build a pre-standup hook that prints the digest before you sit down. Have CI annotate a deploy with a check-in entry.

Stack it with other MCP servers. Gemini CLI happily talks to several MCP servers at once. Combine Steady with a GitHub MCP, a Linear MCP, or whatever else you’ve registered, and let Gemini reason across the whole picture in one conversation.

A note on scope

MCP support is a Gemini CLI capability — the consumer Gemini app, gemini.google.com, and the various Workspace surfaces don’t currently expose MCP connectors. If you want Steady inside a Gemini conversation today, the CLI is the path. The good news is that the CLI is free, open-source, and installs in under a minute.

Installation

The Steady MCP server URL is https://app.steady.space/mcp.

From your terminal:

gemini mcp add --transport http steady https://app.steady.space/mcp

Start a Gemini CLI session and run /mcp auth steady to authenticate. Gemini opens your browser, you sign into Steady, and tokens are written to ~/.gemini/mcp-oauth-tokens.json. After that, Steady is available in every session you start. Step-by-step instructions live in the AI Assistants doc.

Security

The connection authenticates via OAuth 2.1 and is scoped to your individual Steady login, mirroring the permissions you already have inside the product. Tokens stay on your machine — there’s no shared service account, and a teammate setting up their own Gemini CLI doesn’t change anyone else’s access.

About Gemini

Gemini is Google's family of AI assistants, available across the web, mobile, Workspace, and Gemini CLI. MCP support currently lives in Gemini CLI — the terminal-native, open-source interface engineers reach for — so that's where Steady plugs in. Once configured, Gemini CLI can read and write your Steady data alongside whatever else you've connected.

Frequently asked questions

Why Gemini CLI specifically and not the Gemini app?

Today, the Model Context Protocol is supported on the CLI side of Google’s AI surfaces. The consumer Gemini app, the web experience, and Workspace don’t yet expose MCP, so any “Gemini ↔ Steady” workflow runs through Gemini CLI in your terminal. We’ll expand this page if and when that changes.

What does it take to wire it up?

One command. Run gemini mcp add --transport http steady https://app.steady.space/mcp to register Steady, then start a Gemini CLI session and run /mcp auth steady to complete the OAuth handshake. Steady’s tools are then available in every subsequent conversation.

What's a good first thing to try?

Pipe last week’s commits into Gemini and have it cross-reference what shipped against the Goal Story updates Steady has on file. Or kick off a session with “give me the team digest, then summarize what’s blocked and where I can help” — Gemini pulls the digest from Steady and reasons over it without you leaving the shell.

Can I use this in scripts and automations?

Yes. Gemini CLI is scriptable, so anything you can ask it interactively you can also bake into an automation — a nightly job that pulls Goal Story progress and drops a markdown summary somewhere, a pre-standup hook that fetches today’s digest, or a post-deploy step that posts a check-in noting what went out.

How are credentials handled?

Authentication uses OAuth 2.1 against your individual Steady account, so Gemini sees the same data you can see — nothing more, nothing wider. Gemini CLI stores the resulting tokens locally at ~/.gemini/mcp-oauth-tokens.json; nothing is shared across teammates or machines.