> **Source:** https://knowledge.leegality.com/deal-collaboration/Obligations/obligations-in-processing-hub
> **Site:** Leegality Knowledge Base — https://knowledge.leegality.com
> **About:** Leegality is a document execution platform covering eSigning, stamps, NeSL, workflows, and REST API integration.
> **Navigation:** Every article on this site has a plain-text version at `<article-url>.txt` (this format). To get an index of all articles with their `.txt` links, read: https://knowledge.leegality.com/llms.txt
> **AI Guide:** For instructions on how to navigate this knowledge base as an AI agent, read: https://knowledge.leegality.com/ai-readable.txt

---
# Obligations in the Processing Hub

> When the AI reads a contract for obligations, it does that work in the background rather than while you wait. Each contract gets its own extraction job, and the

When the AI reads a contract for obligations, it does that work in the background rather than while you wait. Each contract gets its own extraction job, and the **Obligations** tab in the Processing Hub is where you follow those jobs. You can see which contracts have been read, which are still waiting, which failed and why, and which are ready for you to review.

This is the page to come to when obligations you expected have not appeared. It tells you whether the contract was never read, was read and produced nothing, or ran into a problem.

The other tabs in the Processing Hub track batches of uploads. The Obligations tab is different, because it shows one row for every contract rather than one row for every batch.

> **Info — Note**
>
> A contract can only be read for obligations once it has been executed. Until then, no extraction job exists for it and it does not appear on this page.

## Opening the Obligations Tab

1. Go to the **Processing Hub**.
2. Open the **Obligations** tab. Every extraction job is listed, with the most recent first.

Use the search box to find a job by contract name, and use **Filters** to narrow the list by status, by whether extraction was automatic or manual, by counterparty, or by when the job was queued or extracted.

## Columns

| Column | What it shows |
| :--- | :--- |
| **Document Name** | This column shows the contract that the job ran on. You can click the name to open the contract. If the contract is no longer available to you, the column reads **Document unavailable**. |
| **Counterparty** | This column shows the counterparty on the contract. It is hidden until you turn it on through **Configure Columns**. |
| **Queued At** | This column shows when the job was accepted. The list is sorted by this column with the newest first, and it is the only column you can sort on. |
| **Extracted At** | This column shows when extraction finished. It stays empty while the job is still running. |
| **Extraction Source** | This column shows whether the job started on its own or because someone asked for it. It reads **Auto Triggered**, or **Manually Triggered** with the person's name below it. |
| **Status** | This column shows where the job has reached. The statuses are explained below. |
| **Obligations Found** | This column shows how many obligations were created against how many templates applied to the contract, written as `3/5`. It shows a dash when no template applied. |

You can hide any column except Document Name, Queued At and Status.

## Extraction Statuses and What You Can Do

<!-- ARCADE · P2 · Track an extraction job
     Show: the Obligations tab with rows at Queued, Processing and Completed, and the Obligations Found count.
-->

A job moves through these states on its own. It starts as **Queued**, moves to **Processing** when a worker picks it up, and ends as **Completed**, **Failed** or **Skipped**. **Discarded** is the exception, because it only happens when someone cancels the job.

What you can do to a job depends on where it has reached, and on your own permissions. An action may not be offered even when it is listed below.

| Status | What it means | What you can do |
| :--- | :--- | :--- |
| **Queued** | The job has been accepted and is waiting its turn. Nothing has been read yet. | • **Refresh** brings the row up to date.
• **Discard** cancels the job before it runs. |
| **Processing** | The contract is being read now. The Obligations Found count can rise while the job sits in this state. | • **Refresh** brings the row up to date.
• **Discard** cancels the job before it finishes. |
| **Completed** | Every template that applied to the contract was evaluated and the job has finished. | • **Review** opens the obligations that were found. It reads **View Obligations** when every applicable template produced one.
• **View failure reasons** appears if anything was recorded along the way.
• **Re-extract** reads the contract again. |
| **Failed** | One or more templates could not be extracted or saved. A reason is recorded for each one that failed. | • **Re-extract** reads the contract again.
• **View failure reasons** explains what went wrong. |
| **Skipped** | The job ended without reading the contract, for example because no template applied to it. | • **Re-extract** reads the contract again.
• **View failure reasons** explains why it was skipped.
• **Discard** removes it from your active list. |
| **Discarded** | Someone cancelled the job, or it was skipped because an earlier run had already extracted obligations from this contract. Discarded jobs stay in the list so that you can see what happened. | • **Re-extract** reads the contract again.
• **View failure reasons** explains why it was discarded. |

A status of **Completed** means the reading has finished, not that anyone has checked the result. Reviewing and verifying happen on the obligation itself. See [Adding an Obligation](https://knowledge.leegality.com/deal-collaboration/Obligations/add-obligation).

**Review** and **View Obligations** both take you to the contract's detail view, where the obligations are listed against the clauses they came from.

**View failure reasons** records a reason for each template that failed, so a single job can list several. Discarded jobs use the same place to explain why they were discarded.

> **Caution — Important**
>
> Re-extracting replaces the obligations already extracted from that contract. Anything a person edited or verified on those obligations is replaced along with them.

### Discarding a Job

Discard a job when you do not want the contract read, for example because it was uploaded by mistake or is not one you need to track.

1. Select the entry in the list.
2. Click **Discard**.
3. Confirm when you are asked:

    > **Discard extraction job**
    > Are you sure you want to discard the extraction job for `{name}`? It will be marked as Discarded in the Processing Hub.

The row is not removed from the list. It stays with a **Discarded** badge, and the reason is recorded with its failure reasons, so there is always a record of the decision.

## Downloading a Report

Click **Download report** to export the list of extraction jobs. You can download every job, or only the jobs that match the filters you have applied.

## Related Pages

- [What is an Obligation?](https://knowledge.leegality.com/deal-collaboration/Obligations/obligations-overview)
- [Adding an Obligation](https://knowledge.leegality.com/deal-collaboration/Obligations/add-obligation)
- [Obligation Settings](https://knowledge.leegality.com/deal-collaboration/Obligations/Settings/obligation-settings)

<!-- ==================== POC REVIEW NOTES (not rendered) ====================
This page was rewritten against the implemented frontend slice
(slices/processing-hub/: obligations-columns.js, obligations-cta-buttons.js,
obligations-table.js, translations-en-us.js) and the backend extraction filter
config. The previous version was pattern-matched from the Contracts Processing
Hub and its four statuses (Pending / Ready for Review / Completed / Failed)
did not exist — the real set is Queued / Processing / Failed / Completed /
Skipped / Discarded. The old "Draft — Needs Verification" caution has been
removed since the content is now code-derived, but POC confirmation is still
needed on the following:

1. ACTION PLACEMENT. Deliberately not stated. The feature spec says up to two
   inline CTAs per row (maxActions={2}); the shipped table passes
   maxActions={0}, which would push every action into the row's overflow (...)
   menu. Confirm where each action actually appears before adding placement
   detail or screenshots.

2. "SKIPPED" MEANING. The status is mapped through from the extraction service
   and the code carries no user-facing explanation. The meaning given here
   ("outside the scope of every template") is inferred and needs confirming.

3. RE-EXTRACT ON A COMPLETED JOB - ANSWERED. The product's own confirmation
   dialog states it: "This will re-extract obligations from the document.
   Existing extracted obligations will be replaced." So it is replace-all, not
   add-alongside, and it is not a user choice. Documented as a caution.

4. BULK ACTIONS. Not documented. Row selection is disabled in this build
   (selectable={false}), though retry/discard-selected strings and a
   "You can act on up to 100 jobs at a time" message are already translated.
   Confirm whether bulk actions are shipping; if so this page needs a section.

5. MANUAL EXTRACTION ENTRY POINT. Rows can be "Manually Triggered", but where
   a user triggers extraction from is still unconfirmed (same open question as
   on the overview page).

6. PERMISSIONS. Actions are gated by per-row permission flags
   (retry_allowed / discard_allowed / review_obligations_allowed). Which roles
   or entitlements produce them is not documented anywhere yet.

7. DOWNLOAD REPORT FORMAT. The previous version claimed an XLS export with
   "confidence flags". Neither the format nor confidence flags could be
   confirmed, so both claims were dropped. Confirm the file format.
========================================================================== -->