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

# Set up with a coding agent

> Pair your coding agent, install five verified Archal skills, and create a first sandbox.

The fastest setup path is through the coding agent already working in your
repository. It can inspect your provider clients, choose the smallest useful
integration, and keep Archal separate from your harness and eval stack.

## 1. Sign up

Create an Archal account and open
[Dashboard Setup](https://archal.ai/dashboard/setup). New self-serve accounts
receive \$5 in credit after signup. The remaining \$15 arrives when the first
sandbox is ready to use. No card is required, and the credit does not renew.

## 2. Copy the setup prompt

Choose Claude Code, Codex, Cursor, Devin, or the portable skill target. The
dashboard creates a short-lived pairing prompt for that target.

Paste the full prompt into your coding agent. Treat it like a temporary secret.
Do not commit it, put it in an issue, or paste it into CI logs.

The agent will:

1. exchange the pairing code for a protected Archal credential
2. install exactly five verified skills in the repository
3. verify access to the full environment catalog
4. report readiness only after every checksum passes

Successful setup ends with:

```text theme={null}
Archal is connected and ready
```

Setup does not change application code, start a sandbox, or add a CI workflow.

## 3. Ask for one real test

Give the coding agent a concrete workflow. For example:

```text theme={null}
Use Archal to test our GitHub issue triage flow. Show me the five GitHub
starting states, let me choose one, then create a disposable sandbox and point
our existing GitHub client at it. Prove one read, one mutation, the state diff,
reset, and cleanup. Do not change our evaluator or observability setup.
```

The environment choice belongs to you. You can also leave the starting state
blank and ask the agent to help prepare explicit custom state.

## 4. Review what the agent returns

For a useful first run, expect:

* the sandbox ID and selected environment IDs
* the sample ID and SHA-256 hash, or the validated custom state file
* readiness and expiry
* the application configuration changed to use returned provider URLs
* one provider-shaped read and mutation
* an explicit state diff
* reset or teardown confirmation
* redacted credentials

The Archal workspace key belongs only in the control process. Code under test
receives the scoped provider credentials returned for its environment.

## Next

* [How sandboxes work](https://docs.archal.ai/sandboxes/overview.md)
* [All 22 environments](https://docs.archal.ai/environments/overview.md)
* [Add Archal to CI](https://docs.archal.ai/integrations/ci.md)
* [Manual quickstart](https://docs.archal.ai/manual-quickstart.md)


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