Doclium Developer Platform
Agent Manifest OpenAPI 3.1 Open Web App

Document Intelligence & PDF Engine for Agents and Developers

High-speed vector transformations, forensic redaction, structured data extraction, and pre-send quality audits via REST APIs, Model Context Protocol (MCP), and Autonomous Agent protocols.

API v1 Live & Operational
MCP Server: json-rpc-2.0
Zero AI Model Training Guarantee
Base URL: https://doclium.com/api/v1

1. Authentication

Pass your API key in every request using the Authorization or X-API-Key header:

Authorization: Bearer doclium_sk_live_...
X-API-Key: doclium_sk_live_...

Don't have a key? Go to the API Keys Dashboard to mint one in 1 click.

2. cURL Example — Extract Data

Extract structured invoice or bank statement rows:

curl -X POST https://doclium.com/api/v1/pdf/extract-data \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -F "[email protected]"

Python Quickstart

import requests

API_KEY = "doclium_sk_live_YOUR_KEY"
headers = {"Authorization": f"Bearer {API_KEY}"}

# 1. Run Doclium Ready Quality Audit
with open("contract.pdf", "rb") as f:
    resp = requests.post(
        "https://doclium.com/api/v1/pdf/check",
        headers=headers,
        files={"file": f},
        params={"intent": "sign"}
    )
    print("Ready to Sign:", resp.json()["ready"])
    print("Findings:", resp.json()["findings"])

# 2. Extract structured totals and dates
with open("invoice.pdf", "rb") as f:
    data = requests.post(
        "https://doclium.com/api/v1/pdf/extract-data",
        headers=headers,
        files={"file": f}
    ).json()
    for field in data.get("fields", []):
        print(f"{field['label']}: {field['value']} (Page {field['page']})")

API Keys Management

API keys allow programmatic access to Doclium REST endpoints and MCP server. Secret keys are hashed with SHA-256 and only shown once.

Loading your API keys...

Model Context Protocol (MCP) Integration

Doclium implements the open Model Context Protocol (JSON-RPC 2.0). Connect Claude Desktop, Cursor, or Antigravity to empower LLMs with direct PDF transformations and structured document parsing.

Claude Desktop Configuration

Add this to your ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%/Claude/claude_desktop_config.json (Windows): 

{
  "mcpServers": {
    "doclium": {
      "command": "python3",
      "args": ["-m", "server.mcp_server"],
      "cwd": "/path/to/Doclium"
    }
  }
}

Running this way, every tool's document_id is a path to a PDF on your own machine — Claude Desktop is running locally, so a local file is what it can read. Give it an absolute path (e.g. /Users/you/statement.pdf).

Exposed MCP Tools

  • doclium_merge_pdf: Merge multiple PDF documents into one.
  • doclium_split_pdf: Extract page ranges or pull out specific pages.
  • doclium_compress_pdf: Reduce PDF size while keeping text sharp.
  • doclium_extract_data: Extract structured financial tables, totals, and dates.
  • doclium_check_document: Run Doclium Ready pre-send audit (PII, signatures, math).
  • doclium_redact_pdf: Byte-level character removal for court or privacy sanitization.
  • doclium_export_markdown: Convert PDF into clean Markdown with heading hierarchy.

Tools that produce a new PDF (merge, split, compress, redact) write it next to the source file and return that path, alongside the file's contents as base64.

Over HTTP (agents, not a desktop app)

POST /mcp speaks the same JSON-RPC 2.0, authenticated with an API key (X-API-Key or Authorization: Bearer). Here document_id is not a path — it's the id returned by POST /documents after you upload a PDF to your account, since a hosted server has no access to a remote agent's filesystem. Upload first, then pass the returned id to any tool above.

AI Agent Protocols & Discovery

Doclium provides machine-readable manifests so autonomous AI agents can self-discover and bind document tools at runtime.

Agent Discovery Endpoint

For LangChain, AutoGPT, and Antigravity:

GET /.well-known/agent.json

OpenAPI 3.1 Schema

For ChatGPT Custom Actions & Swagger:

GET /api/v1/openapi.json

LangChain Tool Example

from langchain_core.tools import tool
import requests

@tool
def audit_pdf(file_path: str, intent: str = "sign") -> dict:
    """Audit a PDF document before signing or sending using Doclium Ready."""
    with open(file_path, "rb") as f:
        resp = requests.post(
            "https://doclium.com/api/v1/pdf/check",
            headers={"Authorization": "Bearer doclium_sk_live_..."},
            files={"file": f},
            params={"intent": intent}
        )
    return resp.json()
POST /api/v1/pdf/extract-data

Extract structured totals, VAT/tax, dates, invoice numbers, and account details.

curl -X POST https://doclium.com/api/v1/pdf/extract-data \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -F "[email protected]"
POST /api/v1/pdf/check

Run deterministic Doclium Ready pre-send audit (intents: send, sign, pay, submit, file, all).

curl -X POST "https://doclium.com/api/v1/pdf/check?intent=all" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -F "[email protected]"
POST /api/v1/pdf/compress

Compress a PDF file while preserving visual clarity.

curl -X POST https://doclium.com/api/v1/pdf/compress \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -F "[email protected]" \
  --output compressed.pdf
POST /api/v1/pdf/merge

Combine multiple PDFs into a single downloadable PDF.

curl -X POST https://doclium.com/api/v1/pdf/merge \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -F "[email protected]" \
  -F "[email protected]" \
  --output merged.pdf
POST /api/v1/pdf/redact

Permanently delete sensitive terms (comma-separated).

curl -X POST https://doclium.com/api/v1/pdf/redact \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -F "[email protected]" \
  -F "terms=Confidential,4000-1234-5678" \
  --output redacted.pdf

Interactive API Sandbox

Test Doclium APIs directly from your browser.