One agent per employee,
one workspace for the company.

Every employee gets their own agent, with memory, tools, and scheduled tasks. Your team manages agents and shared connections in one workspace.

Illustrative workspace

Set up the team. Not every agent.

Share tools and skills across the workspace. Each employee gets their own agent, conversation, and memory.

A company workspace in OneCLI. A shared layer across the top holds 12 skills, 8 connected MCP servers and 6 CLIs, built once and available to everyone. Below it, maya in finance, raj in engineering and wei in support each have their own agent that inherits the shared layer and adds a personal scope, each in its own VM with its own credentials.

Illustrative workflow

Delegate a task. Follow the work.

It opens the tools, does the work, and stops on its own when a step needs a human.

One ask from maya in Slack goes to her own agent running in a sealed sandbox with credentials it can never read. On its own the agent pulled 1,204 Stripe charges, matched them against our books in Postgres finding 3 that don't match, and posted the summary to #finance. The $840 refund is held, waiting on kim to approve.

Your tools. More than one way to use them.

Agents can use connected MCP tools, run CLIs, call APIs, or write scripts to complete a task. The gateway applies your access rules to requests routed through it.

MCP tool calls

Connected tools and services

CLI commands

Commands available in the agent environment

curl and raw HTTP

A shell and a URL is still a request

The code the agent writes

Scripts written to complete a task

An MCP gateway governs only MCP tool calls. OneCLI sits at the network layer, so MCP, CLI commands, curl, and generated code all route through the same vault, policies, and audit log.

You decide what needs approval.

Let routine work continue. Pause the actions you choose for approval. See how configured rules allow, block, or hold a request below.

Keys stay in the vault

For connections managed through OneCLI, the gateway adds service credentials to outbound requests. Those credentials are not placed inside the agent environment. Access tokens and other data still need to be protected.

.envproxied by onecli

OPENAI_API_KEY=sk-proj-Xh4mQ2████████f8Kwonecli-managed ✓

STRIPE_SECRET_KEY=sk_live_51Hx8m████████Rq2vonecli-managed ✓

GITHUB_TOKEN=ghp_uV4nR7Tk████████p3Xzonecli-managed ✓

AWS_SECRET_ACCESS_KEY=aK9dPmXw████████L7Rqonecli-managed ✓

DATABASE_URL=postgres://acme:pg4s█████@db.acme.ioonecli-managed ✓

SLACK_BOT_TOKEN=xoxb-8214-Ju7wK████████m2Nponecli-managed ✓

ANTHROPIC_API_KEY=sk-ant-api03-R5kT████████v8Nqonecli-managed ✓

7keys exposed to agents

Want this for your environment?

Open source. Run it on your infrastructure.

Read the code, build your own skills, and self-host OneCLI. Or use the hosted app to get started.

Explore the source →Self-hosting instructions →

Core code is Apache-2.0. Enterprise features have a separate license.

Get started

Give your team their first AI teammates.

Free plan with $5 in AI credits and 500 calls a month. No credit card required.

Create your first agentSelf-host OneCLI