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

# Getting started

> Authenticate, select the correct Vida account, and make a verified API request.

## 1. Create an API token

Create a token at the account scope that should own the integration, then store it outside source
code and prompts. Vida API authentication uses the `token` query parameter. See
[Authentication](/docs/api-reference/authentication) for the scope-specific menu locations.

## 2. Identify the account scope

Most integrations need a stable Agent account ID:

* `targetAccountId` selects the account whose resources the request should use.
* `accountId` owns a newly created Task and normally matches `targetAccountId`.
* `agentConfigId` identifies a staging, live, or saved Agent configuration; it is not an account ID.
* Computer Agent routes place the selected account ID directly in `/computer/accounts/{targetAccountId}`.

Higher-level tokens can use `targetAccountId` for authorized child accounts. Never assume an organization account, Agent account, and Agent configuration share the same ID.

## 3. Identify the key, then read the target

First identify the account that owns the key. Do not include `targetAccountId` in this request:

```bash theme={null}
curl -s "https://api.vida.dev/api/v2/account?token=$VIDA_API_KEY"
```

Use the returned account type and hierarchy to understand the key's scope as described in
[Authentication](/docs/api-reference/authentication). Preserve any `permittedAccountIds` restriction.

If the work concerns another authorized account, read that target separately:

```bash theme={null}
curl -s "https://api.vida.dev/api/v2/account?token=$VIDA_API_KEY&targetAccountId=$TARGET_ACCOUNT_ID"
```

The first response identifies the key owner; the second identifies the selected target. Keep them
separate and confirm both before making changes. For Agent configuration, read staging first. For a
Computer Agent, read its account status and health first.

## 4. Read, change, and verify

Use the same safe pattern for every resource:

1. Read the current state and exact schema.
2. Apply the smallest intended change.
3. Re-read the resource.
4. Verify the real capability, not only the write response.

An accepted lifecycle operation, Task, login flow, or scheduled run is not proof of completion.

## 5. Continue with a workflow guide

* [Configure Agents through the API](/docs/api-reference/agent-guides/overview)
* [Operate Computer Agents](/docs/api-reference/agent-guides/computer-agents)
* [Configure Agent functions](/docs/api-reference/agent-guides/functions)
* [Configure telephony](/docs/api-reference/agent-guides/telephony)
* [Onboard accounts and manage access](/docs/api-reference/platform-guides/accounts-access-and-onboarding)
* [Configure integrations, inbound email, and webhooks](/docs/api-reference/platform-guides/integrations-email-and-webhooks)

For complete examples that combine several resource families, use the API cookbooks:

* [Build and launch an Agent](/docs/api-reference/cookbooks/launch-an-agent)
* [Run an outbound workflow](/docs/api-reference/cookbooks/outbound-workflow)
* [Investigate Agent work](/docs/api-reference/cookbooks/investigate-agent-work)
* [Onboard a reseller customer](/docs/api-reference/cookbooks/onboard-reseller-customer)

Using an AI agent to build or operate your integration? Install the [Vida API Skill](/docs/api-reference/vida-api-skill) so it has the same scoping and verification rules.