Developer docs

The API is available on the Pro plan. Create and manage keys on the API keys page. A key is shown once at creation, so store it securely.

PackForAI ships an open-source MCP server, listed on the MCP Registry. Add it to Claude Desktop (Settings → Developer → Edit Config) or Cursor (.cursor/mcp.json), then ask your assistant to convert a document and it returns clean Markdown inline.

{
  "mcpServers": {
    "packforai": {
      "command": "npx",
      "args": ["-y", "packforai-mcp"],
      "env": { "PACKFORAI_API_KEY": "YOUR_API_KEY" }
    }
  }
}

Published as packforai-mcp on npm. After adding it, try: “Convert ~/Downloads/report.pdf to clean Markdown.”

Convert from your terminal or shell scripts with the zero-dependency CLI. Markdown goes to stdout so you can pipe it; progress and token savings go to stderr.

export PACKFORAI_API_KEY="YOUR_API_KEY"

npx packforai-cli report.pdf                          # print clean Markdown
npx packforai-cli https://example.com/spec.docx -o spec.md
npx packforai-cli scan.pdf --ocr                      # OCR a scanned PDF

Published as packforai-cli on npm. Run npx packforai-cli --help for all options.

Pass your key as a bearer token on every request:

Authorization: Bearer YOUR_API_KEY

Base URL: https://packforai.com/api/v1

Submit a document as multipart/form-data. Fields: file (required), ocr (optional, true/false, Pro OCR for scanned PDFs), webhook (optional, a public https URL notified when the job finishes). Returns a job id.

curl -X POST https://packforai.com/api/v1/convert \
  -H "Authorization: Bearer $PACKFORAI_API_KEY" \
  -F "file=@report.pdf" \
  -F "ocr=false"
# -> { "id": "…", "status": "queued", "statusUrl": "https://packforai.com/api/v1/jobs/…" }

Poll the job until status is completed or failed. On completion you get the compact Markdown inline plus token estimates. Add ?format=full to receive the full document.md (e.g. complete spreadsheet tables) instead of the compact version.

curl https://packforai.com/api/v1/jobs/JOB_ID \
  -H "Authorization: Bearer $PACKFORAI_API_KEY"
# while status is "queued"/"processing", poll again every ~2s.
# when "completed":
# { "id": "…", "status": "completed",
#   "tokens": { "original": 24532, "compact": 8586, "savingsPercent": 65 },
#   "markdown": "# Quarterly Report\n## Financial Summary\n- Revenue: $4.2M …" }

Submit up to 25 files in one request (repeat the file field). Optional ocr and webhook apply to every file. Each file becomes its own job, reported individually.

curl -X POST https://packforai.com/api/v1/batch \
  -H "Authorization: Bearer $PACKFORAI_API_KEY" \
  -F "file=@a.pdf" -F "file=@b.docx" -F "file=@c.xlsx"
# -> { "jobs": [ { "id": "…", "fileName": "a.pdf", "status": "queued" }, … ] }
# Each file is reported individually; jobs count against your plan limits.

Pass a webhook URL to POST /convert or /batch and PackForAI will POST a JSON notification when the job finishes, so you don't have to poll:

// PackForAI POSTs this JSON to your webhook URL:
{
  "event": "job.completed",          // or "job.failed"
  "id": "JOB_ID",
  "status": "completed",
  "tokens": { "original": 24532, "compact": 8586, "savingsPercent": 65 },
  "statusUrl": "https://packforai.com/api/v1/jobs/JOB_ID"
}

Deliveries are signed: verify the X-PackForAI-Signature header with your signing secret (from the API keys page).

import hmac, hashlib

def verify(secret: str, body: bytes, header: str) -> bool:
    expected = "sha256=" + hmac.new(secret.encode(), body, hashlib.sha256).hexdigest()
    return hmac.compare_digest(expected, header)  # header = X-PackForAI-Signature

As a rule of thumb, treat the webhook as a trigger: when it arrives, call GET /jobs/{id} with your key to fetch the result. The URL must be public https.

import time, requests

API = "https://packforai.com/api/v1"
KEY = "YOUR_API_KEY"
headers = {"Authorization": f"Bearer {KEY}"}

# 1. submit a document
with open("report.pdf", "rb") as f:
    r = requests.post(f"{API}/convert", headers=headers, files={"file": f})
r.raise_for_status()
job_id = r.json()["id"]

# 2. poll until completed
while True:
    s = requests.get(f"{API}/jobs/{job_id}", headers=headers).json()
    if s["status"] in ("completed", "failed"):
        break
    time.sleep(2)

# 3. use the clean Markdown
if s["status"] == "completed":
    print(s["markdown"])
    print("saved", s["tokens"]["savingsPercent"], "% tokens")

const API = "https://packforai.com/api/v1";
const KEY = process.env.PACKFORAI_API_KEY;
const headers = { Authorization: `Bearer ${KEY}` };

const form = new FormData();
form.append("file", fileBlob, "report.pdf");

const { id } = await fetch(`${API}/convert`, { method: "POST", headers, body: form }).then((r) => r.json());

let job;
do {
  await new Promise((r) => setTimeout(r, 2000));
  job = await fetch(`${API}/jobs/${id}`, { headers }).then((r) => r.json());
} while (job.status === "queued" || job.status === "processing");

if (job.status === "completed") console.log(job.markdown);

• Pro plan required; usage counts against your plan (100 conversions/day).
• API uploads up to 30 MB per file. For larger files, use the web app.
• Formats: pdf, docx, pptx, xlsx, csv, json.
• OCR (ocr=true) covers scanned/broken PDFs within your monthly OCR page cap.
• Documents and outputs are stored in the EU and never used to train AI.