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

# Obversa

> @obversa/obversa: one dependency that installs the runtime, the runner, the contracts, the workflows, the surfaces, the memory adapters and every engine adapter except Jev, Devin, Mastra and OpenAI Agents.

`@obversa/obversa` is the bundle: one dependency that installs every public
package except the Jev, Devin, Mastra and OpenAI Agents engines, so a
project gets the runtime, the runner, the contracts, the process tools, the
ready-made workflows, the surfaces, the memory adapters, webhook
notifications and the engine adapters at once.
It has no API of its own; import each tool from its own package.

## Install

<CodeGroup>
  ```bash npm theme={null}
  npm install @obversa/obversa
  ```

  ```bash pnpm theme={null}
  pnpm add @obversa/obversa
  ```
</CodeGroup>

The install adds one dependency to your project:

```json package.json theme={null}
{
  "dependencies": {
    "@obversa/obversa": "^0.1.0"
  }
}
```

## Quickstart

After the bundle is installed, import from the member package you need. The
smallest run imports two names from the runtime:

```ts examples/one-agent-job.ts (excerpt) {1-2} theme={null}
import { agentJob, run } from '@obversa/runtime';
import { MockEngine } from '@obversa/runtime/testing';

const engine = new MockEngine(() => 'ready');
```

Every package on the [API reference](/docs/packages) page says "Included in
`@obversa/obversa`" where the bundle installs it, so you can also install
any one of them on its own.

## What it installs

| Package | Page |
| - | - |
| `@obversa/runtime`, `@obversa/runner`, `@obversa/api`, `@obversa/core` | [Runtime](/docs/packages/runtime), [Runner](/docs/packages/runner), [API](/docs/packages/api), [Core](/docs/packages/core) |
| `@obversa/builtin-workflows` | [Built-in Workflows](/docs/packages/builtin-workflows) |
| `@obversa/surface`, `@obversa/surface-diff` | [Surface](/docs/packages/surface), [Surface Diff](/docs/packages/surface-diff) |
| `@obversa/engine-claude-cli`, `@obversa/engine-codex-cli`, `@obversa/engine-grok-cli`, `@obversa/engine-opencode-cli` | [Claude CLI Engine](/docs/packages/engine-claude-cli), [Codex CLI Engine](/docs/packages/engine-codex-cli), [Grok CLI Engine](/docs/packages/engine-grok-cli), [OpenCode CLI Engine](/docs/packages/engine-opencode-cli) |
| `@obversa/engine-claude-agent-sdk`, `@obversa/engine-anthropic-api` | [Claude Agent SDK Engine](/docs/packages/engine-claude-agent-sdk), [Anthropic API Engine](/docs/packages/engine-anthropic-api) |
| `@obversa/memory-git`, `@obversa/memory-simple`, `@obversa/memory-markdown` | [Git Memory](/docs/packages/memory-git), [Simple Memory](/docs/packages/memory-simple), [Markdown Memory](/docs/packages/memory-markdown) |
| `@obversa/notify-webhook` | [Webhook Notifications](/docs/packages/notify-webhook) |

## Jev, Devin, Mastra and OpenAI Agents are separate

The [Jev API Engine](/docs/packages/engine-jev-api) isn't in the bundle. It
needs its own endpoint and key, so install it when you use it:

```bash Terminal theme={null}
npm install @obversa/engine-jev-api
```

The [Mastra Engine](/docs/packages/engine-mastra) isn't in the bundle either.
It runs an agent you built with Mastra, so install it beside Mastra:

```bash Terminal theme={null}
npm install @obversa/engine-mastra @mastra/core
```

The [Devin CLI Engine](/docs/packages/engine-devin-cli) isn't in the bundle
either. Install it beside the Devin CLI:

```bash Terminal theme={null}
npm install @obversa/engine-devin-cli
```

The [OpenAI Agents Engine](/docs/packages/engine-openai-agents) isn't in the
bundle either. It runs an agent you built with the OpenAI Agents SDK, so
install it beside the SDK:

```bash Terminal theme={null}
npm install @obversa/engine-openai-agents @openai/agents zod
```

## Next steps

* [Installation](/docs/get-started/installation): set a project up and run the
  first file.
* [API reference](/docs/packages): every export a user calls, and the page that
  teaches it.


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