Parse CVs in n8n
Use one HTTP Request node: POST the CV file (binary, PDF or DOCX)
as the multipart field file to https://api.verbatimcv.com/v1/parse with
Authorization: Bearer <key>. You get one JSON object back, so the next nodes can
read candidate.full_name or branch on preferences.work_arrangement.value
directly, with no PDF Extract node and no screening prompt to maintain.
Ready-made workflow
Download verbatimcv-cv-triage.json and import it (Workflows → Import from File, or Import from URL). It watches Gmail for emails with a PDF or DOCX attachment, parses the CV with VerbatimCV, and splits candidates into "open to remote" and "onsite, hybrid or not stated", each with the name, the quote and what to confirm. Add your Gmail credential and a Bearer Auth credential holding your VerbatimCV key, then connect the two outputs to Slack, Airtable or your ATS. Tested end to end on n8n 2.46.
Why not PDF Extract plus an LLM node?
That is the usual first build, and it breaks in predictable ways. Two-column CVs (Canva, Novoresume)
come out of plain text extraction as interleaved half-sentences, names in a sidebar go missing, and
the model fills gaps with plausible guesses. VerbatimCV detects columns before extraction, parses
dates in code, and returns null when the CV doesn't say something.
Node settings
- Method: POST. URL:
https://api.verbatimcv.com/v1/parse - Authentication: Generic Credential Type → Bearer Auth, with your key (
vcv_...) as the token. - Body: Send Body on, Body Content Type Form-Data, parameter type n8n Binary File, name
file, Input Data Field Nameattachment_0for a Gmail trigger with attachments downloaded (ordatafor most other nodes). - Timeout (under Options): 60 seconds or more. Most CVs take 5–10 seconds; scans that need OCR take longer.
If the CV is already at a public link, send a form field url instead of file.
Branching on the result
Each preference is {value, source, confidence, evidence}. In an IF node, check
{{ $json.preferences.work_arrangement.source }} is not absent before
using value, and route anything listed in
{{ $json.agent_guidance.confirm_before_acting }} to a human step. The
evidence field is the exact quote, which is useful in a Slack or Airtable summary.
Try it
Get a free API key (50 parses a month, no card), then:
curl https://api.verbatimcv.com/v1/parse \
-H "Authorization: Bearer $VERBATIMCV_API_KEY" \
-F file=@cv.pdf
Agents can use the parse_resume tool instead: uvx verbatimcv-mcp. See the
quickstart for Python, JavaScript and MCP setup.