Upload a SmartSign <BOX> XML — we render it through the
SmartSign API, run a multimodal critic on the rendered image, and save
every revision to History so you can compare before / after.
- Render
- Vision
- Critique
- Apply
Original
v0Edited
rev-001Revisions
What I see in the rendered image
—
Reading the XML and deciding fixes…
—
Issues
Edits
Edits below were applied automatically by the iteration loop.
Flagged
Not auto-patchable — requires manual review.
No sessions yet — upload an XML in the Analyze tab to get started.
—
Original (v0)
XML
Vision findings
Critique result
Edited
XML
What was suggested vs. applied
Drag the slider to wipe between versions. The edited XML is left of the divider; the original XML is right of it.
Line-level diff between Original XML and the selected revision's XML.
LLM spend by session, model, and step
Every vision and critique LLM call is logged. Pricing is best-effort —
override via SMARTSIGN_PRICING_JSON env to match your
current provider rates.
Per-model breakdown
| Model | Calls | Input | Output | Reasoning | Cost |
|---|
Per-step breakdown
| Step | Calls | Input | Output | Cost |
|---|
Sessions ranked by spend
Rows show how many vision/critique calls one upload's auto-iterations consumed.
| Session | Iterations | Vision | Critique | Retry | Total tokens | Cost |
|---|
Last 50 LLM calls
| When | Step | Model | Session · iter | In | Out | Reasoning | Cost |
|---|