> ## Documentation Index
> Fetch the complete documentation index at: https://developers.argosidentity.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Editing a workflow
> How to change playbook actions, step engines, reference notes, and the output schema, and what saving does.
**Where:** workflow detail → **Edit Workflow** (or Actions → Edit Workflow in the list)
**Everything except the policy and the AI model can be changed.** That covers the information section, the playbook actions, the step engines, the output schema, and the reference notes — and none of those edits regenerate the playbook.
## What is on the screen
| Section | Editable | Contents |
| ------------------ | --------- | --------------------------------------------------------- |
| Information | Yes | Name and description |
| Playbook ID | Read-only | Copy button only |
| Playbook Actions | Yes | Action order, name, description, and assigned engines |
| AI Model | Read-only | The model chosen at creation |
| JSON Output Schema | Yes | Also offers a copy button |
| Reference Notes | Yes | The reference notes of every action, merged into one list |
## The policy and the AI model cannot be changed
**There is no policy definition section on the edit screen.** Changing the policy means rebuilding the playbook, the actions, and every engine assignment — which effectively makes it a different workflow. So the policy is fixed after creation, and only the execution plan beneath it is adjustable. The AI model is read-only for the same reason.
You can read the original policy text in `policyText` from [`GET /workflows/:workflowId`](/en/omni/api-reference/get-workflow).
When you do need to change the policy itself, **create a new workflow.** [Copying](/en/omni/dashboard/workflow-setup/workflow-detail#copying) the existing one first is faster, because you do not have to rewrite the output schema.
Check the **reference notes** before you rewrite the policy. A verdict that comes out wrong is usually not caused by the policy sentence itself but by that sentence being carried into the action's reference notes incorrectly. Reference notes can be fixed right on this screen.
## Editing the information section
Only the name and description change; the playbook is untouched.
## Editing playbook actions
This is the edit you will use most. Selecting **Edit** switches the section into edit mode in place, not in a separate window. Change the order, names, descriptions, and engines, then **Save** to apply everything at once.
| Operation | How |
| ---------------- | ---------------------------------------------------------------------------------------------------------------------------------------------- |
| Reorder | Drag with the **drag handle** on the left of the row. Step numbers are reassigned from the top |
| Rename | The **pencil icon** next to the action name. Rules: **starts with a lowercase letter · lowercase letters, digits, and underscores (`_`) only** |
| Edit description | Select the description line and type |
| Change engines | The **n engine(s)** button on the right of the row (**Select engines** when none are assigned) |
| Delete an action | The **trash icon** on the right of the row |
| Add an action | **+ Add Action** at the bottom of the list |
**Save** is disabled while nothing has changed. **Cancel** returns everything to how it was before the edit.
| Error | Cause |
| ------------------------------------------------------------------------------------------------ | -------------------------------------------------------- |
| Must start with a lowercase letter and use only lowercase letters, digits, and underscores (`_`) | The name rule was broken |
| Duplicate action name | Two or more actions share a name |
| Some actions cannot be saved | Save was attempted with either error above still present |
`step` is only the execution order, so reordering reassigns the numbers. When your integration code refers to a step, use `workId` — it stays fixed regardless of order.
## Editing step engines
Selecting **n engine(s)** (or **Select engines**) on the right of an action row opens a window for choosing the engines assigned to that step.
* The title is **Edit Step n Engines**.
* There is a **Select All** checkbox and a count such as "2 of 7 selected".
* Each engine shows its name and description, so you can confirm what it does right here.
* **Apply** puts the choice on that step; the final save happens with **Save** on the playbook actions section.
Engine capabilities, prices, and supported countries are in [AI model and engines](/en/omni/dashboard/workflow-setup/engine-selection).
## Editing the output schema
An **Edit Output Schema** window opens with the same two tabs used at creation, **Field Builder** and **JSON Schema Input**.
| Element | Description |
| --------------------------- | ----------------------------------------------------------------------------------------------- |
| Drag handle | Reorders fields |
| Expand arrow | Opens and closes the child properties of `object` and `array` fields |
| Number badge next to a name | How many child properties that field has |
| **+** button | Adds a child property to an `object` or `array` field |
| Trash icon | Deletes a field |
| Format/Constraint | Format for `string` (None, Date, Email, and so on); maximum size for `array`, such as `Max: 10` |
The constraints match creation — the root must be `"type": "object"`, nesting is 2 levels at most, and names cannot repeat. The full rules are in [output schema](/en/omni/dashboard/workflow-setup/output-schema).
Changing the schema does **not** change the results of analyses that already finished. An analysis freezes and stores the schema as it was at run time. The new schema applies from the next analysis onward.
## Editing reference notes
This is the first place to look when an action carried the policy's intent across incorrectly. The **Edit Reference Notes** window lists the entries in a text area as lines beginning with `- `.
| Key | What it does |
| ----------- | ------------------------------------------- |
| Enter | Creates a new `-` item |
| Shift+Enter | Continues the current item on the next line |
| Limit | Value |
| ---------------- | -------------- |
| Entries | 20 per action |
| Length per entry | 500 characters |
Saving **replaces the whole list**. **Save** is disabled while nothing has changed.
## What saving does
| What you changed | Playbook regenerated | `playbookId` | `playbookVersion` |
| ------------------------------- | -------------------- | ------------ | ----------------- |
| Information (name, description) | No | Unchanged | Unchanged |
| Playbook actions | No | Unchanged | Increments |
| Step engines | No | Unchanged | Increments |
| Reference notes | No | Unchanged | Increments |
| Output schema | No | Unchanged | Increments |
`playbookId` stays the same for the life of the workflow; only `playbookVersion` goes up with each edit. Past analyses froze the playbook as it was at run time, so edits do not affect them — [reading analysis results](/en/omni/guides/reading-analysis-results#which-playbook-produced-this-verdict).
With nothing to change, the screen tells you there are no changes to save.
## Common errors
| Message | Meaning |
| ----------------------------------------------------- | -------------------------------------------------------------------- |
| This workflow has no playbook | Playbook generation has not finished, or it failed |
| Step not found | The action was deleted elsewhere before you saved. Refresh and retry |
| Engine already assigned | The same engine was assigned twice on one step |
| This engine is in use by a step and cannot be removed | Some step still uses it. Unassign it in the step engine editor first |
| No access | The account does not have permission to edit this workflow |
## Related pages
* [Creation complete and workflow detail](/en/omni/dashboard/workflow-setup/workflow-detail) — copying, deleting, and the profile table
* [Policy writing guide](/en/omni/guides/policy-writing-guide) — for when the policy itself has to be rewritten