An API for engineering drawing data
SpecCapture's REST API takes a drawing file and returns each sheet's title block, revision history, parts list and notes as JSON, with a confidence score, a review state and a bounding box on the page for every value. You send the file, then check the job or receive a signed webhook when it is done, and download the results as JSON, CSV or Excel.
How it works
With SpecCapture, step by step.
Get a key
Create an API key in the SpecCapture app.
Send the file
POST it to /v1/extractions. The answer comes at once with a job; the reading runs in the background.
Wait for it
Check the job, or add a webhook and receive a signed message when the file is finished.
Read the values
Fetch each sheet's values, or download the whole job as JSON, CSV or an Excel workbook.
curl https://api.speccapture.com/v1/extractions \
-H "Authorization: Bearer $SPECCAPTURE_API_KEY" \
-H "Idempotency-Key: $(uuidgen)" \
-F "file=@bracket.pdf"The API docs: a quickstart, code samples and the full reference
What to check
Before you rely on it.
Respect the review state
Each value says whether it was accepted automatically, is waiting for review, or was confirmed or corrected by a person. A new workspace starts with every value waiting, so import only what your process allows.
Retry safely
Send an Idempotency-Key with each upload: a retry with the same key never creates, or charges for, a second job.
Expect to wait
Each sheet is read on its own, in the background: on our test drawings a sheet took 36 to 49 seconds on average.
File limits
Up to 50 MB and 200 sheets a file, and a sheet up to 150 million pixels. Accepted: PDF, TIFF, PNG, JPEG, BMP and WebP.
Every fact in one place, with how accuracy was measured: SpecCapture at a glance.
Questions
What people ask.
Does SpecCapture have an API?
Yes: a REST API with keys you create in the app. It takes the same files as the app and returns the same values, with webhooks for finished files.
Can an AI assistant use SpecCapture?
Yes. SpecCapture also works from AI assistants such as Claude, GitHub Copilot in VS Code and Cursor: its MCP server at https://api.speccapture.com/mcp lets them send drawings and read their values, with the same API keys. Each one's setup is at https://speccapture.com/docs/ai-assistants/ in the API docs.
Is there an engineering drawing OCR API that gives confidence scores and bounding boxes?
Yes. SpecCapture reads each sheet's printed text with OCR, which works on scans as well as PDFs, has an AI model pick out each field's value, and boxes each value where its text was found on the sheet. So rather than the page's words, the API returns each named value, such as the part number or a parts-list cell. Through the API, each value comes with a confidence score from 0 to 1, a review state and, while the drawing is kept, a bounding box in page coordinates from 0 to 1, measured from the top left of the sheet.
What does the API return?
For each sheet, the title block's values and every table row, each with its value as it stands, the value as first read, a confidence score from 0 to 1, a review state and a bounding box in page coordinates.
Does the API answer straight away?
It answers the upload at once with a job. Each sheet is read on its own, in the background: on our test drawings a sheet took 36 to 49 seconds on average.
How is the API priced?
The same as the app: each sheet uses one of the workspace's drawing sheets, from its plan or a credit pack. Starter is $79 a month for 100 drawing sheets; Pro $199 for 350, Business $499 for 1,100 and Scale $999 for 3,750. Credit packs need no plan: 60 sheets for $60, 275 for $250 or 700 for $600, each one payment. A pack's sheets count for 12 months, and a workspace on a plan uses its plan's sheets first.
Early access
Be first to try SpecCapture.
Leave your email and we will let you know when early access opens. Plans start at $79/mo.
We use your address only to tell you about SpecCapture. Privacy