Nimbus
Your 24/7 Employee.
An AI that logs into your apps, sends emails, updates your CRM, and runs tasks while you sleep. Set it up once, free forever.
Built by Sunday Johnson at Yoodule.
Two ways to use Nimbus
Self-host
Free forever
- Runs on your laptop, your server, your CI
- One-command install on macOS, Linux, Windows
- Full source, BUSL-1.1 license
- SHA256-verified, SLSA-attested releases
Nimbus Cloud
Managed · Yoodule-operated
- We provision, patch, and monitor 24/7
- Production-grade uptime and incident response
- Migration help from self-hosted when you're ready
- Same Nimbus, same MCP servers, no ops burden
Self-hosting Nimbus is and always will be free. Cloud is for teams who'd rather pay us to run it than to run it themselves.
Why Nimbus
You don't want another tool. You want an employee — one that handles the busywork, learns how you work, and is there at 2 AM when a customer emails. Nimbus is that employee. It plugs into the apps you already use, follows the steps you'd take, and gets smarter every time it runs.
| Works while you sleep | Schedule a task at any time — daily follow-ups, weekly reports, hourly checks. Nimbus runs them and pings you when it's done. |
| Uses the apps you already pay for | Notion, Gmail, your CRM, your calendar. Nimbus logs in and acts on them — no copy-paste, no exports. |
| Asks before it does anything risky | Sends a draft instead of an email, shows you a plan before it deletes a row. You're always in control. |
| Heals itself when something breaks | A password rotated, a service restarted, an app updated — Nimbus notices and fixes it. You don't get paged. |
What it can do
📨 Send follow-up emails
Tell Nimbus "email everyone who downloaded the whitepaper last week but didn't book a call." It writes a draft, waits for your sign-off, and sends.
📊 Build reports from your data
"Pull last month's closed deals, group by industry, and put a chart in the shared Notion page." Done by morning. No spreadsheet wrangling.
🗓️ Triage your inbox
Every morning, Nimbus labels new emails, drafts replies for the obvious ones, and flags the rest in Slack. You only look at the ones that need you.
🔁 Run workflows on a schedule
"Every Friday at 4pm, post the week's top wins to #team-channel and email me a summary." Nimbus runs it forever, no maintenance.
🧠 Remembers how you work
The more you use Nimbus, the more it learns. It knows which emails you always archive, which reports matter, and which steps to skip.
🛡️ Never sends anything you haven't approved
Nimbus drafts first, you sign off. It can't send an email, delete a row, or post a message without your click. Full audit trail, every action.
Where it fits in your business
For a founder doing everything
Stop spending Sundays writing follow-up emails and updating your CRM. Hand the repetitive work to Nimbus and use the time for the stuff only you can do.
nimbus start — set up once on your laptop, free forever.
For a small team
Give everyone on the team their own Nimbus employee. They share context, remember what each person prefers, and never lose a customer note.
Every action is signed, audited, and reversible.
For a company that needs control
Run Nimbus inside your own infrastructure, with your own keys and your own audit logs. Or hand it to us and we run it. Same employee either way.
The Semantic Gateway (mcp.json)
For the curious: Nimbus runs on a small config file that lists every app your employee can use. You probably won't need to touch it — Nimbus comes with sensible defaults.
Nimbus aggregates all your stdio and HTTP-based MCP servers into a single interface. When an AI model requests a capability, Nimbus dynamically discovers and executes the appropriate tool using vector embeddings stored in Qdrant.
Configure your active servers by modifying mcp.json in your Nimbus home directory:
{
"mcpServers": {
"playwright-mcp": {
"url": "http://localhost:3100/mcp",
"transport": "streamable-http"
},
"notion-mcp": {
"url": "https://mcp.notion.com/mcp",
"transport": "streamable-http",
"auth": "oauth"
},
"linkedin-mcp": {
"command": "uv",
"args": ["run", "-m", "linkedin_mcp_server", "--transport", "stdio"]
}
}
}
The gateway exposes four main endpoints to models (under the namespace nimbus-utility-mcp):
list_servers— Lists all configured MCP backend servers with their connection status and tool count.find_tools— Searches across all servers semantically. Your model asks for "browse linkedin" and gets only the relevant tool schemas.execute_tool— Executes any tool on any registered server dynamically.chain_tools— Chains multiple steps together sequentially (e.g. search → read → send) in a single API roundtrip.
CLI Reference
For the technical user: every command Nimbus understands. Most people only need nimbus start and nimbus doctor — the rest is for when something goes sideways.
The nimbus CLI controls the lifecycle of your local containerized workspace.
| Command | Action |
|---|---|
nimbus start |
Boot services. Provisions and starts all Docker containers (postgres, redis, qdrant, gateway). |
nimbus start --build |
Re-build & boot. Rebuilds Docker images from the current source and boots services cleanly. |
nimbus start --gateway |
Headless mode. Runs only the gateway container — skip the full stack for CI or remote MCP-only use. |
nimbus start --fresh |
Wipe data volumes. Regenerates Postgres/Redis credentials and re-initializes the data directory. |
nimbus stop |
Tear down. Safely terminates all container services and frees system ports. |
nimbus upgrade |
Upgrade. Pulls the latest binaries and upgrades the wrapper setup. |
nimbus update |
Update the CLI. Refreshes the gateway container image (with --gateway). |
nimbus uninstall |
Uninstall. Removes the Nimbus CLI, containers, and configuration. |
nimbus uninstall --keep-data |
Uninstall but preserve databases, keys, and configuration. |
nimbus uninstall --purge |
Uninstall and wipe everything without prompting. |
nimbus doctor |
Diagnose. Checks Docker, version skew, container health, update freshness, credential drift. |
nimbus recover-pg-role |
Repair. Migrates the nimbus_postgres_data volume to the current .env credentials. |
nimbus recover-redis-password |
Repair. Rotates the Redis volume password to match the current .env. |
nimbus dashboard install |
Install the optional Agent Dashboard companion. |
nimbus dashboard uninstall |
Remove the Dashboard container and service block. |
nimbus dashboard status |
Show whether the Dashboard is enabled and its container state. |
nimbus config <subcommand> |
Manage ~/.nimbus/.env (list, get, set, unset). |
nimbus mcp <subcommand> |
Manage ~/.nimbus/mcp.json (list, get, set, remove). |
nimbus chat "prompt" |
Send a prompt to the dashboard and stream the response. |
nimbus env-init |
Populate CLI-owned secrets in ~/.nimbus/.env (idempotent). |
See nimbus --help for the full reference, or the Download page for platform-specific install instructions.
Ready to give your agent a real runtime?
Want a guided setup? Schedule a 1-on-1 onboarding session and we'll walk through your workspace.
Schedule Onboarding Session →