> ## Documentation Index
> Fetch the complete documentation index at: https://docs.getcargo.ai/llms.txt
> Use this file to discover all available pages before exploring further.

> ## Agent Instructions
> If you are an AI coding agent operating Cargo, install the agent skills first: paste `Install Cargo by following https://api.getcargo.io/INSTALL.md` into your session, or run `npx skills add getcargohq/cargo-skills`. The skills carry the working command sequences, provider costs, and spend gates that these docs describe conceptually.
> The skills index at https://raw.githubusercontent.com/getcargohq/cargo-skills/main/llms.txt lists every skill, recipe, and provider playbook individually. Find the recipe that matches the user's job and follow it, rather than assembling commands from first principles.
> Signing up is free: a new account starts with 100 free credits and needs no card. `cargo-ai login --email you@company.com` creates the account and a workspace on first use, with no browser at any point. Tell the user this before asking them to sign up.
> Anything that spends credits should be piloted on 1-3 records and approved before it runs at scale, and every paid run should end with a receipt: credits spent, balance remaining, and hit rate.

# Replicas

> Launch background coding agents in cloud workspaces

Replicas runs coding agents in cloud workspaces built from an environment — a config bundle that carries the repository binding plus the variables, files, skills and MCPs the agent gets. The Cargo integration lets a workflow launch an agent and use whatever it produced (branches, pull requests) in downstream nodes.

## Authentication

Connect to Replicas using an API key created from the Replicas dashboard under **Organization → Settings → API Keys**.

### Connection details

| Field | Description |
| - | - |
| **API Key** | Your Replicas API key |

***

## Replicas actions

### Launch agent

Create a workspace and send an initial message to a coding agent.

**Configuration**

| Field | Description |
| - | - |
| **Environment** | The Replicas environment the workspace is created from |
| **Prompt** | The initial message sent to the coding agent |
| **Workspace name** | Name of the workspace — whitespace is replaced with dashes |
| **Coding agent** | Claude Code, Codex, Cursor, Opencode or Pi |
| **Model** | Model identifier for the selected coding agent |
| **Thinking level** | How much reasoning effort the agent puts in |
| **Plan mode** | Have the agent propose a plan instead of editing code |
| **Compute size** | Small (2 vCPU, 8 GB) or Large (4 vCPU, 16 GB) |
| **Lifecycle** | Whether the workspace keeps running, sleeps or is archived once the agent is done |

The **Environment** dropdown lists the team and personal environments in your Replicas
organization. The Global environment is not listed: it only layers configuration onto the
other environments and has no repository binding, so it cannot back a workspace.

Leave **Workspace name** empty to get a name derived from the workflow node and the record
being processed.

**Waiting for the agent**

The node stays pending until Replicas calls back, so downstream nodes run against a
finished agent rather than a workspace that has only just started. Replicas notifies Cargo
directly — nothing is polled — and the node resumes on the `replica.turn_completed`
callback. A `replica.error` callback fails the node with the reason Replicas reported.

**Output**

| Field | Description |
| - | - |
| `id` | Workspace identifier |
| `name` | Workspace name |
| `status` | Workspace status when the turn finished |
| `url` | Link to the workspace in the Replicas dashboard |
| `branches` | Branch, default branch and pull request URLs per repository |
| `pullRequestUrls` | Every pull request URL the agent opened during the turn |

***

## Use cases

<AccordionGroup>
  <Accordion title="Automated code changes" icon="code-branch">
    Turn a qualified request into a branch and a pull request without a human in the loop.
  </Accordion>

  <Accordion title="Codebase research" icon="magnifying-glass">
    Ask an agent to investigate a repository and feed its findings into the rest of the workflow.
  </Accordion>

  <Accordion title="Triage at scale" icon="list-check">
    Launch one agent per incoming issue and collect the pull requests they open.
  </Accordion>

  <Accordion title="Scheduled maintenance" icon="clock">
    Run dependency bumps or migrations on a schedule and review the resulting pull requests.
  </Accordion>
</AccordionGroup>

***

## Best practices

1. **Give the agent a complete prompt** - It cannot ask follow-up questions mid-run
2. **Pick the environment carefully** - It decides which repository the agent works on
3. **Use plan mode to review first** - Have the agent propose an approach before it writes code
4. **Sleep or archive when done** - Workspaces bill per minute while they stay up


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.