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

# Build accountable work with OrgX

> Use OrgX from the AI tools you already work in, or build accountable work into your product with the OrgX API.

OrgX keeps your team's decisions and the work your AI tools do in one place,
with a link to where each came from. Connect it to ChatGPT, Claude or another
AI tool and ask what needs your attention; the next chat starts from what is
already decided.

## Use OrgX with your AI tool

Follow one path from connection to a result you can check.

**1. Connect one tool**

In ChatGPT, follow the [ChatGPT guide](https://docs.useorgx.com/docs/guides/chatgpt-integration.md): open
the OrgX listing and choose Connect. No terminal needed. For Claude,
Cursor, Codex and others, use [client setup](https://docs.useorgx.com/docs/guides/toolkit.md).

**2. Get your first useful result**

Ask: "Use OrgX to show my connected workspace and summarize what needs my
attention. Include links to the records. Don't change anything." You should
see your workspace and what is waiting on you, with links. A new workspace
may have nothing waiting yet. If the answer never mentions OrgX, start a
new chat and name OrgX in the prompt.

**3. Review work**

Open a linked decision or piece of work in OrgX. Work can be checked
against your [quality bar](https://docs.useorgx.com/docs/guides/quality-bar.md) (beta); accepting it
stays your call.

**4. Continue in another tool**

Ask a second connected tool about the same record. It starts from the
decision instead of from zero. Beta.

**5. Know the limits**

See [availability by client](https://docs.useorgx.com/docs/resources/roadmap.md) for what each tool can
read, change and recover today, and what to do when something fails.

> [card] [Connect my tool](https://docs.useorgx.com/docs/guides/toolkit.md)
>
> Pick ChatGPT, Claude, Cursor, Codex or another client.

> [card] [See an example first](https://docs.useorgx.com/docs/guides/synthetic-data-trial.md)
>
> Explore prepared example data without installing or connecting anything.

> [card] [Delegate a small outcome](https://docs.useorgx.com/docs/quickstart.md)
>
> Save one bounded draft, choose access and inspect the first result.

## Build with the API

Create work from your product and keep the receipt for every operation.

> [card] [API quickstart](https://docs.useorgx.com/docs/api/quickstart.md)
>
> Authenticate, create work and read the operation receipt.

> [card] [API reference](https://docs.useorgx.com/docs/api/overview.md)
>
> Authentication, operations, errors and SDKs.

To try the API without an account, validate a portable receipt. The request
does not store the submission:

```bash theme={"dark"}
curl -s https://useorgx.com/agent-work-receipts/examples/codex \
  | curl -s https://useorgx.com/api/v1/agent-work-receipts/validate \
    -X POST \
    -H 'Content-Type: application/json' \
    --data-binary @-
```

The response reports whether the receipt is valid, what outcome its producer
reported, and what evidence it carries.


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