POST /v1/sites/:id/approve-research

Approve a site that’s paused in research_review and advance it into generation. Optionally merge final edits — services, hours, FAQs, photos, brand colors — into the research payload before the build runs. Only relevant for sites created with "review": "manual"; auto-mode sites advance themselves and never sit in research_review.

Cost: Free (the generation itself still costs the usual ~200–500 credits — see POST /v1/sites).

Request

curl -X POST https://api.warpweb.ai/v1/sites/8f3c2a1b/approve-research \
  -H "Authorization: Bearer wwk_<your-key>" \
  -H "Content-Type: application/json" \
  -d '{
    "services": [
      { "name": "Drain cleaning", "description": "Hydro-jetting and cabling.", "price": "from $189" },
      { "name": "Water-heater install" }
    ],
    "hours": {
      "Monday": "8:00 AM - 6:00 PM",
      "Sunday": "Closed"
    },
    "excludeImages": ["research/photo-3.jpg"],
    "ownerPrompt": "Emphasize 24/7 emergency service"
  }'

All body fields are optional. An empty {} is valid — it accepts the research payload as-is and starts the build.

Body

Field	Type	Description
`services`	array	Replace the services list. `{ name, description?, price? }`. Empty `name` entries are dropped.
`testimonials`	array	Replace the testimonials list. `{ quote, author?, rating?, source? }`. Empty `quote` entries are dropped.
`faqs`	array	Replace the FAQ list. `{ question, answer }`. Empty `question` entries are dropped.
`hours`	object	Weekday → hour string (e.g. `"Monday": "9:00 AM - 5:00 PM"` or `"Closed"`). Replaces the hours map wholesale.
`brandColors`	object	`{ primary?, secondary?, accent?, personality? }`. Hex strings. Omitted slots preserve research-extracted values.
`existingSiteUrl`	string	Override the existing-site URL recorded at create time.
`ownerPrompt`	string	Voice / tone direction merged into the build prompt. Same field as on create.
`businessDescription`	string	One-sentence pitch the generator uses for the hero / meta-description.
`designStyle`	string	Override the auto-selected design style. Unknown values are silently ignored — the existing style stays.
`allowStockPhotos`	boolean	Toggle the stock-photo (Pexels) layer of the photo cascade. Default `false` as of 2026-05-24 — sites with empty upload + Google Places slots render Lucide icons over brand colors instead of generic stock imagery. Set `true` to opt in to stock fallback when you’d rather have generic photo coverage than icon-and-color sections. Also accepted at create time on `POST /v1/sites`.
`excludeImages`	string[]	Drop photos from generation. Each entry can be either the `localPath` (for `existing_site` / `upload` / `facebook` photos: e.g. `"images/existing-3.jpg"`) or the full `storage_url` (for `places` / `stock` photos, which only live in the photo catalog and not on disk: e.g. `"https://...supabase.../places-XYZ.jpg"`). Use whatever the `photos[]` entry from `GET /v1/sites/:id/research` exposed — both forms are accepted. Excluded photos are tombstoned in the photo catalog (`metadata.excluded=true`), filtered out of the build picker, AND removed from the on-disk staging dir when local. Surfaces on subsequent `GET /research` calls with `exclude: true` so your UI can pre-check the operator’s prior exclusions.
`uploadedPhotos`	string[]	URLs to download and add to the photo library. Same field as on create.
`contentPlan`	object	Partial overrides to the AI-generated content plan (e.g. `servicesToHighlight`, `competitiveAdvantages`). Merged shallow over the existing plan.
`contactEmail`	string	Operator-confirmed contact email. Written to `sites.contact_email` before the build kicks off. Use this when the email cascade returned null at research time (`email_status: "missing"`) or when the operator wants to swap the placeholder for the customer’s real address. After the site is built, use `POST /v1/sites/:id/contact` instead. See Email resolution.

services, testimonials, faqs, and hours write into the priority-1 content source — they beat anything the extractor finds during research.

Response

{
  "success": true,
  "phase": "generating"
}

The build runs asynchronously from here. Listen for site.complete (or site.failed) on the customer lifecycle webhook, or poll GET /v1/sites/:id until status flips to complete.

Errors

Status	Body	Cause
400	`{ "error": "Site is not in research review phase" }`	The site has already advanced (or never entered) `research_review`. Check `generation_phase` on `GET /v1/sites/:id`. Idempotent — repeated calls after a successful approve also land here.
400	`{ "error": "No research data found" }`	The site exists but `research_data` is empty. Should not happen for sites that reached `research_review`; usually indicates a corrupted row.
404	`{ "error": "Site not found" }`	No site with that ID belongs to your account.

GET /v1/sites/:id/research — fetch the payload you’re approving.
PATCH /v1/sites/:id/research-draft — save in-progress edits without advancing the phase.
POST /v1/sites — set "review": "manual" to land here.

GET /v1/sites/:id/research PATCH /v1/sites/:id/research-draft

POST /v1/sites/:id/approve-research

Request

Body

Response

Errors

Related