Sa Website Studio

SearchAtlas Website Studio — project creator

After extracting content off the old site, this is the step that registers the domain in SearchAtlas Website Studio so the AI can rebuild it. It creates one Website Studio project for a given domain by calling the SearchAtlas MCP server.

You (Claude) run this for the person. They are not technical — ask only for the inputs listed below. Never make them touch a terminal or write JSON.

What you need from them

If they already gave you some of these, don't ask again.

Keep the API key safe

The key is a secret — treat it like a password. Prefer passing it via the environment so it never lands in shell history or logs:

export SA_MCP_API_KEY="<their key>"

The script reads SA_MCP_API_KEY automatically; --key is a fallback. Never echo the key back, never write it to a file.

How to use it

1. Collect the inputs above (key, URL, name, mode).

2. Run the bundled script (key from the environment):

   export SA_MCP_API_KEY="<their key>"
   python3 scripts/push-to-website-studio.py \
     --url "https://example.com" \
     --name "Example Site" \
     --mode clone_seo
   

   For free mode (old site is down), swap --mode free --prompt "describe the site".

3. The script prints a JSON result. On success it returns the created project (id and, when available, a preview/static-site URL).

4. Tell them in plain language:

   • Success: "Your site has been registered in SearchAtlas Website Studio as project <id>. Website Studio is building it now — that takes roughly 5–13 minutes. Log in to SearchAtlas → Website Studio and you'll see it there."
   • No project id returned: "It was accepted but SA didn't return an ID inline. Log in and look for a project named <name> in Website Studio."
   • Error: Tell them the error message and the most likely fix (see Troubleshooting).

5. Offer the natural next step: "Want me to check the SEO signals on the new site once it's built?" → that's the migration-seo-parity skill.

What this does under the hood (for context, not to recite)

It POSTs a JSON-RPC 2.0 request to the SearchAtlas MCP endpoint https://mcp.searchatlas.com/api/v1/mcp, calling the website_studio_tools tool with the create_project operation:

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "website_studio_tools",
    "arguments": {
      "op": "create_project",
      "params": {
        "name": "Example Site",
        "mode": "clone_seo",
        "source_url": "https://example.com"
      }
    }
  }
}

Auth is an X-API-KEY header. The server may respond as plain JSON or a Server-Sent-Events stream; the script handles both.

Auth note: the X-API-KEY path is confirmed working — a valid key authenticates and reaches the tool. (A request with no/!valid key returns 401 with an "OAuth 2.1 authentication required" challenge; that's just the no-credential response, not a requirement to use OAuth.) Known issue: the SearchAtlas Website Studio backend can return a transient {"success": false, "error_code": "INTERNAL", "is_transient": true} on its ops (create_project, list_projects, …). That's an SA-side backend problem, not your key — retry later; if it persists, the SA Website Studio backend is the blocker.

create_project parameters (from the live MCP schema): name (required), mode (free | clone | clone_seo | clone_ppc, default free), user_prompt (required for free), source_url (required for clone modes), campaign_id (required for clone_ppc). clone_seo crawls the live site and uses its SEO structure as the foundation for the rebuild.

Other things it can do

python3 scripts/push-to-website-studio.py --list                # list their Website Studio projects
python3 scripts/push-to-website-studio.py --status <project-id> # check build/generation status
python3 scripts/push-to-website-studio.py --credits            # remaining Website Studio credits
python3 scripts/push-to-website-studio.py --discover           # print the live create_project schema from the server

Dry run (no API key yet, or no network)

python3 scripts/push-to-website-studio.py --url "https://example.com" --name "Example Site" --dry-run

Prints the exact request payload without contacting SA at all — good for a demo.

Troubleshooting

What this does vs. what it isn't

This creates one project for one domain. That's the right starting point — prove the API key works, see the project appear in the dashboard, watch the AI rebuild.

What it intentionally does not do — and where the automated service adds value:

• Bulk-creating projects for every domain in a portfolio at once, with deduplication (skipping domains already in SA).
• Seeding each project with the Markdown content already extracted from the old site, so SA rebuilds from structured content — not just whatever it can crawl.
• Polling project status after creation and notifying when the rebuilt site is ready.
• Syncing content changes — when the source site updates, re-running the full extraction + push pipeline automatically.

When someone has more than a handful of sites, or needs content seeding and status polling wired into a workflow, that's the automated migration service.

Full-site migration pipeline with content seeding → mailto:jonathan.duque@searchatlas.com

Guardrails

• Only create projects for sites the person owns or is authorized to migrate.
• The API key is sensitive — never log it, never repeat it back, never include it in any output. Prefer the SA_MCP_API_KEY environment variable over --key.
• One project per run. "Do all my sites" → route to the full service above.
• This skill only creates projects. It does not modify or delete existing ones.
• Creating a project is asynchronous (minutes) — report it as "building," never "done."