First data room with a full API, CLI & MCP

Build virtual data rooms with code.

REST API. Native CLI. Model Context Protocol server. Type-safe SDKs. The developer reference for programmable data rooms. For humans and agents.

Full docs Quickstart ↗Browse the API

Three surfaces, one platform

Pick the integration shape that fits your runtime.

The REST API is the canonical surface. The CLI wraps it for shell pipelines. The MCP server wraps it for AI agents. Same auth, same scopes, same data model.

REST API

HTTP + JSON + bearer tokens

A deliberately boring REST surface. 43 operations across 6 resources. Stable contracts. S3 presigned URLs for uploads.

curl https://api.papermark.com/v1/datarooms \
  -H "Authorization: Bearer $PAPERMARK_TOKEN"

Read the rest api guide

CLI

npm install -g papermark

Scripted workflows and CI. JSON output. Dry-run mode. Built for cron jobs, GitHub Actions, and local exploration.

papermark datarooms create --name "Q4 Investors" \
  --json | jq -r '.data.id'

Read the cli guide

MCP server

Agent-native control

Drop @papermark/mcp-server into Claude Desktop or any MCP host. 43 tools. Real API calls. No mocks.

npx -y @papermark/mcp-server
# Authenticated via PAPERMARK_TOKEN

Read the mcp server guide

Why this exists

Data rooms have been point-and-click for 30 years.

Every other developer surface. Payments, email, infrastructure, search. Went programmable. Data rooms didn't. We're fixing that.

Programmable by default

Create a dataroom, attach documents, mint signed links, and pull analytics. All from code. No drag-and-drop. No human-in-the-loop required.

Agent-ready

Built for Claude, ChatGPT, and the agent stack. MCP server ships 43 tools. Auth scopes constrain what agents can do.

Auth that scales

Dashboard tokens for CI you own. OAuth 2.1 device flow with PKCE for tools you ship. Scopes restrict per-token blast radius.

Open source core

Read the full case for developer data rooms

60-second tour

From token to shared dataroom in five calls.

One curl creates the room. One uploads a doc. One mints a link. One pulls analytics. The fifth is optional.

1. Authenticatebash

export PAPERMARK_TOKEN=pm_live_…
curl https://api.papermark.com/v1/me \
  -H "Authorization: Bearer $PAPERMARK_TOKEN"

2. Create a dataroombash

curl -X POST https://api.papermark.com/v1/datarooms \
  -H "Authorization: Bearer $PAPERMARK_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"name": "Series A. Acme"}'

3. Attach a documentbash

curl -X POST https://api.papermark.com/v1/datarooms/$ID/documents \
  -H "Authorization: Bearer $PAPERMARK_TOKEN" \
  -F "file=@deck.pdf"

4. Mint a tracked linkbash

curl -X POST https://api.papermark.com/v1/links \
  -H "Authorization: Bearer $PAPERMARK_TOKEN" \
  -d '{"dataroom_id": "'$ID'", "password": "pelican-42"}'

Full quickstart on papermark.com/docs

For agents

The data room your agent can actually drive.

dataroom.dev is written for two readers: developers and language models. Every page exposes machine-readable equivalents.

/llms.txt, concise site index for crawlers and agents
/llms-full.txt, full content flattened to markdown
JSON-LD, schema.org metadata on every page
OpenAPI, machine-readable API contract
MCP, direct tool access for hosts
robots.txt, GPTBot, ClaudeBot, PerplexityBot all allowed

How to build agents that operate datarooms

Start building

Get a token. Make a call. Ship in an afternoon.

Free tier on Papermark. Open source self-host available. Bring your own agent.

Read the docs →Quickstart ↗