Skip to main content
Version: v2026.0.0

Managing scripts in the admin UI

The Scripts section of the admin panel lets administrators author and deploy custom front-end JavaScript through a governed lifecycle that includes an LLM-based security scan. A typical use is the host connector that wires Quick Prompt into a business application: instead of hand-deploying a static .js file, the connector is authored in the admin panel, scanned, and served by uxopian-ai to authenticated users. The same mechanism carries the quick AI actions of the IBM Content Navigator plugin and the AI menu of the ARender viewer.

Scripts are managed per tenant. Changes take effect without restarting the application.

The Scripts section of the admin panel listing scripts with their status

Figure: the Scripts section lists each script with its publication status (here, a force-published script and a draft). The list is paginated and filtered on the server (search, status).

The script lifecycle​

Figure: A script is created, edited as a draft, scanned, and — once certified or force-published — deployed and served.

A script always has two faces:

  • Draft content — what the administrator is editing. Not servable to users.
  • Deployed content — the last published version. This is what end users receive.

Create a script​

  1. In the admin panel, open Scripts and click New Script.
  2. Enter a name. The name is the script's ID: it is immutable and may contain only letters, digits, _, and -. A name already used by another script is reported while you type, and Save is refused.
  3. Save. The script is created empty; add its content on the detail page.

Edit a script​

Open a script to reach its detail page, which provides a code editor and the script's current status.

The script detail page with the code editor, status badge, and Scan & publish actions

Figure: the script detail page — code editor, status badge, and the Delete / Discard / Save / Scan & publish actions.

ActionDescription
SaveSaves the editor content as the draft. Editing the draft of a certified script invalidates its certification — it must be re-scanned.
DiscardReverts the editor to the last saved/deployed state.
Scan & publishRuns the security scan; on a Certified verdict, the draft is automatically published and becomes servable.
Force publish(In the Scan & publish dropdown.) Deploys the current draft without scanning. Use with care — the action is logged and attributed.
DeleteRemoves the script after confirmation.
Content is edited through the draft only

You cannot modify a script's deployed content directly. All edits go through the draft; publishing promotes the draft to the deployed content.

Status​

StatusMeaning
CertifiedThe latest scan certified the draft as safe
RejectedThe latest scan rejected the draft; the scanner's comment is shown
Force publishedThe deployed content was published without a scan
DeployedThe script has deployed content and is being served
Pending scanTransient indicator shown while a scan is in progress

The security scan​

The scan sends the draft content to a configured LLM with a security-review prompt and expects a verdict of Certified or Rejected with a comment. This requires an LLM provider to be configured for scanning:

SettingEnvironment variableDefaultDescription
script.scan.llm-provider-idSCRIPT_SCAN_LLM_PROVIDER(empty)ID of a configured LLM provider used to run the scan. Required to enable scanning.
script.scan.llm-modelSCRIPT_SCAN_LLM_MODEL(empty)Model to use; falls back to the provider's default model when blank
script.scan.prompt—(built-in)The security-review prompt sent to the model

If no scan provider is configured, script create/edit/delete and force publish still work, but the Scan & publish action is unavailable. See Configuration — script-scan.yml.

Serving a script​

A deployed script is served to any authenticated user as application/javascript, with HTTP caching (Cache-Control and ETag, so unchanged scripts return 304 Not Modified):

GET /api/v1/scripts/{id}

The ETag is a hash of the script's ID and its deployed content, so a browser picks up a new deployment instead of keeping the old script on 304. A script that has never been deployed is not servable and returns 401 Unauthorized. A draft can be previewed (for development) via GET /api/v1/scripts/{id}/draft.

To load a deployed script in a host page:

<script src="https://your-gateway/api/v1/scripts/myConnector"></script>

REST API endpoints​

Administration (/api/v1/admin/scripts):

MethodEndpointDescription
GET/api/v1/admin/scriptsList scripts — a page { collection, total, facets } of summaries (without the code: derived status and line count). Filters: search, status, workspaceId; pagination: offset, limit, orderBy (id, not name), ascending
GET/api/v1/admin/scripts/{id}Get a script by ID
GET/api/v1/admin/scripts/availability?id={id}Whether a new script can take this ID (true / false)
POST/api/v1/admin/scriptsCreate a script from its ID only — body {"id": "my-script"}; 409 if the ID exists
PUT/api/v1/admin/scripts/{id}Update the draft content
POST/api/v1/admin/scripts/{id}/scanRun the security scan
POST/api/v1/admin/scripts/{id}/publishPublish a certified draft
POST/api/v1/admin/scripts/{id}/force-publishPublish the draft without scanning
DELETE/api/v1/admin/scripts/{id}/draftDiscard the draft
DELETE/api/v1/admin/scripts/{id}Delete the script (204 No Content)

Serving (authenticated users):

MethodEndpointDescription
GET/api/v1/scripts/{id}Serve the deployed script as application/javascript (401 if never deployed)
GET/api/v1/scripts/{id}/draftServe the draft as application/javascript (for preview)