Alpheric AI for the terminal
Atlas CLI brings agents, workflows, automation, and project intelligence to the command line — so developers and technical teams can run their entire AI stack without ever leaving the shell.
npm i -g @alpheric/cliYour AI stack, one command away
Authenticate once, then run agents, trigger workflows, and inspect projects with fast, scriptable commands that fit right into your shell.
$ atlas login
Opening browser to authenticate...
Logged in as [email protected] (workspace: acme-prod)
$ atlas agents run growth-report
Running agent "growth-report"...
[routed] atlas-reason steps: 4 tokens: 12.4k
Report ready -> ./out/growth-report.md
$ atlas workflows trigger qbr --account acme
Triggered workflow "qbr" (run_a91f3c)
status: running stage: gather-context (2/6)AI shouldn’t live in a browser tab
Technical teams move fastest in the terminal. Everything else just gets in the way.
Context-switching
Jumping between dashboards, tabs, and consoles breaks flow and slows every task.
Glue scripts
Brittle one-off scripts to wire models, data, and tools together pile up fast.
Slow automation
Shipping a simple automation means tickets, reviews, and waiting on the dashboard.
Everything you need, in the shell
A focused set of commands that cover the full surface of your AI workflows.
Run agents
Invoke autonomous agents with tools, memory, and grounded context from one command.
Trigger workflows
Kick off multi-step pipelines across models and tools, with arguments and flags.
Inspect projects
Query project state, runs, and intelligence without leaving your terminal.
Sync knowledge
Push and pull documents and data sources so context stays fresh everywhere.
Automate tasks
Script repeatable jobs and chain commands into your existing toolchain.
Connect environments
Switch between workspaces and environments with scoped, isolated credentials.
From login to output in three commands
Authenticate, run an agent, and capture structured JSON you can pipe straight into the rest of your toolchain. No SDK, no boilerplate — just your shell.
# Authenticate once, then automate anything
$ atlas login --workspace acme-prod
# Run an agent and capture structured output
$ atlas agents run support-triage \
--input "March invoice looks wrong" \
--json > triage.json
# Output
{
"run": "run_7c21d0",
"routed": "atlas-reason",
"status": "completed",
"summary": "Billing mismatch flagged for review"
}Up and running in five steps
No setup ceremony. Install, authenticate, and you are issuing commands in minutes.
Install
Add the CLI globally with a single package manager command.
Authenticate
Log in once with a scoped token; sessions are encrypted locally.
Connect workspace
Select the workspace and environment you want to operate against.
Run command
Trigger agents, workflows, and automations straight from your shell.
Review output
Stream results, inspect runs, and pipe output into your own tools.
Safe to run in production
Built so individual developers and large teams can adopt it with confidence.
Scoped tokens
Short-lived, revocable tokens with least-privilege scopes per workspace.
Env isolation
Each environment keeps its own credentials, config, and state -- no leakage.
Audit trails
Every command is logged and attributable, so teams keep full visibility.
// CLI Questions
CLI
Questions
How Atlas CLI brings agents, workflows and organizational knowledge into your terminal.
Talk to our teamAtlas CLI is a developer-first AI command environment. It runs agents, triggers workflows, inspects projects, syncs knowledge and automates tasks directly from the terminal, so developers stay in the environment they already work in.
Developers and platform teams who prefer the terminal to a browser, and who want AI capability inside scripts, pipelines and automation rather than in a separate chat window. It is the programmable surface of the Alpheric intelligence layer.
Running agents on demand, triggering multi-step workflows, inspecting project state, syncing organizational knowledge, and automating repetitive operational tasks. Anything expressible as a command can be scheduled or chained into existing pipelines.
They are two interfaces onto the same intelligence layer and the same organizational knowledge. Atlas App is the visual workspace; Atlas CLI is the terminal equivalent. Work started in one is visible in the other.
// Get Started
Ready to build smarter?
One secure platform for AI models, agents, and enterprise workflows.
Live in days, not months. No vendor lock-in.