Feature
Optional AI brief assist: a design spec, not pixels.
POST /v1/briefs turns a one-sentence brief into an editable /v1/renders request body. The deterministic engine renders it; AI never generates app UI.
AI drafts the spec, not app UI
appscreenshotapi does not fabricate product screens. The renderer composites the real captures you provide, pixel-exact. The brief assist only drafts the design spec (preset choice, theme tokens, headline and caption copy, slide roles) when you call POST /v1/briefs.
The drafted spec is a plain /v1/renders request body. Teams and agents can review, edit, or replace any field before rendering, and the final render is fully deterministic and reproducible.
Your key, your policy
Pass an llm object (provider, model, key) so brief drafting runs through your approved model provider and account instead of a shared vendor key. There is no server-side LLM key, so you pay your own provider, and the key is used once for the upstream call and never stored, cached, queued, or logged.
Briefs can include optional app metadata and captures to slot into the drafted slides, so the spec comes back close to your listing, and the compliance lint report still checks the rendered output either way.
curl https://api.appscreenshotapi.com/v1/briefs \
-H "Authorization: Bearer $ASA_KEY" \
-H "Content-Type: application/json" \
-d '{
"brief": "A calm, dark, premium look for a habit tracker aimed at indie builders",
"llm": {
"provider": "anthropic",
"model": "claude-haiku-4-5",
"key": "sk-ant-..."
},
"app": { "name": "Acme Habits", "tagline": "Build habits that stick." }
}'API details
Endpoint POST /v1/briefs (optional, off the render path)
Credential llm: { provider, model, key }. Your key, used once, never stored
Output an editable /v1/renders design spec plus a short rationale
Use it in a release pipeline
Render the set from CI with POST /v1/renders, check the compliance lint report, then download the finished images for review before App Store Connect or Google Play upload.