Spin any site into a scraper.

Paste one URL or a batch. Weaver respects robots.txt, discovers the useful fields, verifies the rows, and hands you a scraper that runs without AI.

Enter comma-separated fields. After scaffolding the page, Weaver will show recurring fields and other useful fields worth requesting.
Scaffold the page to reveal recurring and recommended fields.
Try 20
✦ Guide Weaver describe the records you want

Choose a starter or describe your own dataset. Weaver ranks same-site sections and GET searches allowed by the active server policy, verifies a repeated collection, then separates fields it found from useful fields still worth requesting. Guidance applies to every URL in a batch.

Dataset starters

Robots policy shown per run · private networks blocked · API key stays server-side

Rows extracted
0
Scrapers woven
0

The idea

AI builds it. Python runs it.

Ready to weave. Drag Weaver around the screen, use arrow keys to move, or press Home to return Weaver to the silk thread.

Live pipeline · ready

your-site.com

The loom

The loom

Watch the weave.

Weaver checks permission on every page, maps a representative record, verifies the generated Python, then follows same-origin next links until the data ends, repeats, or reaches your safety caps.

  1. Robots
  2. Fetch
  3. Find target
  4. Analyze
  5. Generate
  6. Verify / repair
  7. All pages
  8. Export
Pick a site above, or paste your own

Schema — what Weaver tied down

Nothing woven yet. Every field Weaver tags on the page lands here, tied to its selector and type.

Recommended for this category

Auto-detected suggestions will appear here.

spec.yml — the whole scraper, declared

# written as Weaver ties fields down

Pipeline log

idle

scraper.py — live output

waiting for generated Python

The bolt

Fresh off the loom.

Every passing run keeps its proof: the spec, deterministic scraper, cached fixture, QA report, data exports, and optional correlated images.

Extracted rows
Live preview

No rows yet. Spin a scraper above and the validated output lands here, straight from the run.

Scraper QA feedback

Waiting for a verified scraper.

Woven files

A passing run writes the spec, scraper, fixture, QA report, every export format, and a portable bundle.

AI designs the validated spec once. The generated Scrapling runtime is plain Python and preserves the run's explicit robots policy.

scraper.py — woven, not written


  






  

    
The shuttle

    
Ask for exactly what you need.

    
Give Weaver a public page and a small field contract. The response is the verified Python scraper. If that runtime later breaks, it can report the failure, trigger a rebuild, and refresh your portal with the replacement run.

  

  

    

      

        POST
        /api/runs → /latest
      


      
      

        
        
      


      

        

          Fields to return
          
Name the values your UI needs. Hints can describe custom fields.

        

        
      

      

        
        
        
        
        
      

      

      

      

        Advanced request options
        

          
          
        

        
        
        
      


      

        
        
      

      
      
One public collection page · active robots policy is reported in logs · private networks blocked. Runtime failures send only a bounded error summary—never page HTML—and each scraper artifact may request one replacement run.

    


    

      

        
Generated runtime · scraper.py

        ready
        Open full API docs ↗
      

      

        Fetch
        Find
        Extract
        Validate
        Ready
      

      

        
        
        
        
        
        
      

      
Run a test to weave scraper.py.

The code shown here will be the actual verified Python artifact—not a mock response.

      

        
Portal-ready latest value endpoint

        
      

    

  


  

    

      

        
Developer portal preview

        
Card Sale Watch

        
Illustrative Pokémon card-sale layout. Values always come from the public URL you run above.

      

      

        Illustrative portal
        
        
      

    

    

      

        

          
Card Drop Watchsale tracker · developer GUI

          Waiting for verified data
        

        

          
NO DATARun the scraper—Your verified fields will populate here.

          
NO DATAWaiting for item 2—Last-known-good values stay visible during repair.

          
NO DATAWaiting for item 3—The replacement run updates all cards together.

        

      

      

        
Failure → rebuild → rerunv—

        
systemWaiting for the first verified scraper.