Command Center Artifacts
Working with saved Command outputs, previews, reports, and code snippets
Artifact Actions
- Preview/Edit: opens the artifact in a formatted preview by default. Switch to Edit to change the saved artifact text.
- Pin to Command context: the plus button stores the artifact in Command workspace memory. This only makes the artifact reusable in later Command turns; it does not publish, download, or save it to Knowledge Base. Available to full admins only — scoped User-role logins (Client View Tools) get the same client-facing, read-only workspace view as portal clients and won't see this button.
- Download: downloads the current artifact body with an extension based on its content format.
- Save to Knowledge Base: stores the artifact as a client Knowledge Base doc with type
artifact. This makes the artifact visible in the client's Knowledge Base list instead of only inside the chat. - Mark applied: marks the artifact status as
applied. It is a lifecycle label for operator tracking; it does not publish the artifact by itself.
Formats
Artifacts support these stored kinds:
| Format | Description |
|---|---|
document | General document content |
markdown | Markdown formatted text |
html | HTML with sandboxed preview |
jsx | React component code |
diff | Unified diff output |
preview | Rendered preview content |
report | Structured report data |
asset | Binary or media asset reference |
verification | Verification result notes |
table | Tabular data |
The viewer also detects fenced markdown, html, jsx, tsx, and json blocks. Markdown renders as formatted text. HTML renders in a sandboxed preview and shows the source. JSX/TSX renders as code and downloads as .jsx.
Generated Images
Image generation tool results (image_create_session, image_continue_session) render as their own inline card, separate from the artifact system above — a pending skeleton while the generation runs, then a result card, or an error card if it failed. This renders the same way in the client portal as it does in admin.
Clicking a result opens a version drawer with every image generated in that image session across the chat, prompt/model/provider metadata, download, copy-prompt, and a "New version" action that prefills a continue-session follow-up. Generated assets are saved to the client's asset library automatically.