Connect Cline to Enrichr
Reverse-engineer any website — design system, animations, WebGL/shaders, 3D scene graph, recovered source — directly from Cline's chat. Two minutes, no code.
1Get an API key
Keys look like enrichr_7f3a9c2e1b8d4a6f…. Sign in and create one from your account page — it's shown once, so copy it right away.
2Add the MCP server
Click the MCP Servers icon in Cline's panel → Remote Servers tab, and fill in the name and URL — no JSON to write. Or edit the config directly:
{
"mcpServers": {
"enrichr": {
"type": "streamableHttp",
"url": "https://enrichr.dev/mcp",
"headers": {
"X-API-Key": "enrichr_your-key-here"
},
"disabled": false,
"autoApprove": []
}
}
}Config file location: ~/.cline/mcp.json (CLI), or via Configure MCP Servers in the panel's Configure tab (IDE extension).
"type": "streamableHttp" explicitly. Omitting it defaults to the legacy sse transport, which still works but is slower to connect.Raw HTTP (any client)
# List every tool
curl -X POST https://enrichr.dev/mcp/tools \
-H "X-API-Key: enrichr_your-key-here"
# Call one
curl -X POST https://enrichr.dev/mcp/invoke \
-H "Content-Type: application/json" \
-H "X-API-Key: enrichr_your-key-here" \
-d '{"tool":"rebuild_brief","arguments":{"url":"https://bruno-arizio.fr"}}'3Use it
Ask Cline in plain language:
Give me a rebuild brief for https://bruno-arizio.fr
Extract the design system from https://gsap.com
What's in the 3D scene on https://threejs.org/examples/webgl_animation_keyframes.html
Recover the original source for https://myoldsite.comThe tools
Start here
| Tool | What you get | Credits |
|---|---|---|
rebuild_brief | One document written for an LLM to rebuild the site from: tokens, components, exact animation timings, shader GLSL, scene graph, fonts, assets, recovered source — plus a full capture zip | 5 |
extract_site_profile | Everything above as structured JSON | 3 |
Multi-page — everything above runs against one page. For more than one:
| Tool | What you get | Credits |
|---|---|---|
map_site | Every URL a site declares (sitemap/robots, falling back to same-origin links) — discovery only | 1 |
crawl_site | Discovery, then a real capture of up to max_pages of them | 1 |
Focused extraction
| Tool | What you get | Credits |
|---|---|---|
extract_design_system | CSS variables, palette, type scale, spacing, easings, breakpoints, webfonts | 1 |
extract_source_maps | Original TypeScript/JSX recovered from shipped bundles | 2 |
extract_scene_graph | Three.js meshes, materials, textures, lights, camera, renderer config | 1 |
reverse_engineer_animations | GSAP tweens, ScrollTriggers, Framer Motion, Lenis | 1 |
reverse_engineer_webgl | Canvas info, compiled GLSL, 3D model URLs, draw-call stats | 1 |
reverse_engineer_dom | Full DOM tree, headings, landmarks, tables, forms, link inventory | 1 |
reverse_engineer_assets | Every asset with real sizes and MIME types | 1 |
Plus scraping, OCR, screenshot, and analysis tools — POST /mcp/tools lists all of them.
Troubleshooting
| Problem | Fix |
|---|---|
| Server won't connect | Check "type": "streamableHttp" is set and the URL ends in /mcp |
Invalid API key | Keys start with enrichr_. Check for a trailing space |
This API key has been revoked | Issue a new one from the dashboard |
Monthly credit limit reached | The error names your plan and usage; upgrade or wait for the reset |
| Missing tools | Confirm the server shows as connected in the MCP Servers panel |
| Request timeout | Heavy WebGL sites can take 30-60s. Retry, or use a focused tool |
FAQ
Is my data stored?
Extraction results are not persisted. We record which tool you called, when, and the URL — that is what meters your credits. Page content is not retained.
Can I share my key?
No. Credits are per-account; sharing shares your quota. Issue a second key from the dashboard instead — you can revoke it independently.
Why did an extraction cost more than I expected?
Cost is per tool, not per page. rebuild_brief holds a browser for far longer than a token read, so it costs 5. The table above lists every price.