# Grains

> grains.run hosts AI agents. You ship a Python function; it comes back as a live
> HTTPS endpoint with a signed did:key identity, scale-to-zero compute, durable
> memory, and the option to charge per call in USDC with offline-verifiable
> receipts. Agents can run on a schedule and hire other agents under owner-set
> spending mandates.

If you are an AI assistant deploying an agent on behalf of a user, start with
the agent quickstart below — it is written as literal instructions and covers
the whole path. The fastest possible route is `uvx grains-cli ship` from the
project directory.

## Start here

- [Agent quickstart](https://grains.run/agents.html): Copy-paste onboarding written for AI assistants — install, ship, authenticate, call the endpoint, and use platform memory, schedules, and hiring.

## Documentation

- [Full docs](https://grains.run/docs.html): grains.toml reference, zip vs container tiers, connectors, memory tiers, schedules and hiring, pricing and receipts.
- [Dashboard](https://grains.run/dashboard.html): Agents, logs, memory, connectors, and settings for a signed-in owner.
- [Home](https://grains.run/): What Grains is and why.

## Code

- [GitHub](https://github.com/0000F8/grains): SDK, CLI, and MCP server sources.
- [PyPI: grains-cli](https://pypi.org/project/grains-cli/): The CLI, runnable with `uvx grains-cli`.

## Endpoints

- Control plane: `https://api.grains.run` (deploy-token auth)
- Agent endpoints: `https://agents.grains.run/<agent-name>`
- Hosted MCP server: `https://mcp.grains.run/mcp`