New: Steady CLI & agent skills

Over the past few months months we’ve been expanding Steady into tools you already work in: first the MCP server, then an all-new REST API. Today we’re adding two more: the Steady CLI for working with Steady from your terminal, and agent skills that teach your agents how to use Steady, including pulling your team’s live context into their workflow.

The Steady CLI

The CLI is a self-contained binary — no runtime to install, no dependencies to manage — whose commands mirror the v2 API resources. It’s the right tool when you’re already in a shell: checking in between commits, piping work into a script, or giving an autonomous agent a way to report its own work.

The output is human-readable when you run it yourself, and machine-readable when a script or agent reads it instead. Run a command by hand and you get styled output: cards with clickable titles and formatted Markdown, easy to scan. Pipe it, redirect it, or pass --json, and it switches to JSON automatically, byte-for-byte stable so your scripts and agents keep working unchanged.

A few examples of what you can do with it:

• See what’s new in Steady without leaving your terminal: steady digest
• Surface your team’s blockers in a script: steady check-ins --blocked | jq -r '.[].person.name'
• Have a coding agent post its own progress when a PR merges: steady goals <id> goal-updates create @update.json

Getting started

# macOS / Linux / WSL2
curl -fsSL https://cli.runsteady.com/install.sh | bash

# Windows PowerShell
irm https://cli.runsteady.com/install.ps1 | iex

Then steady auth login to sign in through your browser, and steady --help for the full list of commands. The CLI docs cover install, auth, and example usage.

Agent skills

The tools are only half the story. The other half is teaching your agents to use them well, which is what our new skills library is for. Drop these portable skills into your agentic coding setup and your agent knows how to work with Steady. There are four:

• steady-cli: how to drive the steady binary, from install and auth to the command groups and reading and writing data from a shell.
• steady-api: how to build against the v2 REST API, including authenticating with personal access tokens, generating clients from the OpenAPI spec, and the endpoint and schema details.
• steady-updates: how to write a check-in or goal update that reads like a thoughtful teammate wrote it, with the right length, unambiguous people references, scannable formatting, and context beyond the captured activity. Just as handy when you’re checking in through Claude or ChatGPT as when an autonomous agent writes its own.
• steady-context: live work context for your agents. Before it starts, it pulls what your team is working on, what’s in flight, and what’s blocked, and uses it to inform the work. So the agent works from the same picture your team has: it catches when a change overlaps work already underway, steers clear of duplicating something a teammate just shipped, and grounds its decisions in what’s actually happening around the code. It’ll even flag when what you’re about to do cuts against a current goal.

That last one is the real unlock. A coding agent with your team’s context ships the right work the first time, saving you both time and tokens.

Getting started

In Claude Code, install all four as a plugin:

/plugin marketplace add steadyspacecorp/steady-skills
/plugin install steady-skills

Using Codex, Cursor, the Claude apps, or something else? The skills repo has one-command install steps for every host.

See how it all fits together

The CLI and skills join a growing set of ways to build on Steady: the MCP server, the API, webhooks, and our demo playground. To see them side by side and understand which fits which job, visit the agents & developers page. It maps each tool to the patterns it’s best for, so you can pick the right entry point for what you’re trying to do.