About dropscan-mcp
dropscan-mcp connects MCP clients to the Dropscan postal-mail scanning API. It supports reading mailing details and downloading scans, with optional account actions and recipient, forwarding-address, and webhook management protected by two safeguards.
Key features
- 01List scanboxes and mailings
- 02Read mailing details and OCR text
- 03Download PDF, ZIP, and envelope images
- 04Request scan, forward, or destroy actions
- 05Manage recipients and forwarding addresses
- 06Manage webhook rules
Use cases
- 01Review scanned physical mail
- 02Download a mailing's PDF or OCR text
- 03Request forwarding with explicit safeguards
- 04Configure recipients or webhook rules
MCP support
It exposes Dropscan mailing, download, and account-management operations as tools to MCP clients over stdio. Write and action tools require both an enabling environment variable and `confirm: true`.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
uvx dropscan-mcpDROPSCAN_API_TOKENSecretPersonal access token for the Dropscan REST APIDROPSCAN_ALLOW_WRITESSet to 1 to unlock the write and action tools, which also require confirm=true on every callLocal servers need a runtime on your machine (e.g. Node.js 18+). Config file locations differ by client; see your client’s docs.
Tools
Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.
Details
From the project READMEdropscan-mcp
MCP (Model Context Protocol) server for the Dropscan postal-mail-to-email API. Dropscan receives your physical letters at a scanbox; this server lets your MCP client list mailings, download the scans (PDF / ZIP / envelope image / OCR text), and, when explicitly enabled, request actions (scan / forward / destroy) and manage recipients, forwarding addresses and webhook rules.
mcp-name: io.github.bitterdev/dropscan-mcp
Safety first: Dropscan has no sandbox
Every request hits the real production account. Actions are irreversible: a destroyed letter is physically shredded and gone, a forward physically mails the letter and costs money.
Because of that, all write/action tools are gated twice:
DROPSCAN_ALLOW_WRITESenvironment flag. Unless it is set to1/true/yes, every non-GET tool refuses to run. It is disabled by default.confirm: trueparameter. Each write/action tool takes an explicitconfirmargument (defaultfalse) and refuses unless it istrue.
Read-only GET tools are always available and never change state.
Tools
Read-only (always available)
| Tool | Endpoint |
|---|---|
list_scanboxes | GET /scanboxes |
list_mailings(scanbox_id, current_status?, older_than?) | GET /scanboxes/{id}/mailings |
get_mailing(scanbox_id, uuid) | GET /scanboxes/{id}/mailings/{uuid} |
get_mailing_pdf(scanbox_id, uuid, save_path?) | GET .../{uuid}/pdf |
get_mailing_zip(scanbox_id, uuid, save_path?) | GET .../{uuid}/zip |
get_mailing_envelope(scanbox_id, uuid, save_path?) | GET .../{uuid}/envelope |
get_mailing_plaintext(scanbox_id, uuid) | GET .../{uuid}/plaintext |
list_recipients(scanbox_id) | GET /scanboxes/{id}/recipients |
get_recipient(scanbox_id, recipient_id) | GET /scanboxes/{id}/recipients/{id} |
get_forwarding_address(address_id) | GET /forwarding_addresses/{id} |
get_shipment(shipment_id) | GET /shipments/{id} |
list_webhook_rules() | GET /webhook_rules |
get_webhook_rule(rule_id) | GET /webhook_rules/{id} |
The download tools save the asset to disk (default ~/Downloads, override with save_path or the DROPSCAN_DOWNLOAD_DIR env var) and return the saved path, byte size and content type.
Write / action (double-gated: DROPSCAN_ALLOW_WRITES=1 and confirm=true)
| Tool | Endpoint |
|---|---|
request_mailing_action(scanbox_id, mailing_uuid, action_type, forwarding_address_id?, forwarding_date?, confirm) | POST .../{uuid}/action_requests |
cancel_mailing_action(scanbox_id, mailing_uuid, action_request_id, confirm) | DELETE .../{uuid}/action_requests/{id} |
set_mailing_recipient(scanbox_id, mailing_uuid, recipient_id, confirm) | PUT .../{uuid}/recipient |
create_recipient(scanbox_id, type, firstname?, lastname?, name?, confirm) | POST /scanboxes/{id}/recipients |
create_forwarding_address(firstname, lastname, street, number, zip_code, city, country_code, company?, info?, state?, confirm) | POST /forwarding_addresses |
create_webhook_rule(url, events, authorization_header?, confirm) | POST /webhook_rules |
update_webhook_rule(rule_id, url?, events?, authorization_header?, confirm) | PATCH /webhook_rules/{id} |
delete_webhook_rule(rule_id, confirm) | DELETE /webhook_rules/{id} |
action_type is one of scan, forward, destroy (exact enum from the API). For forward, both forwarding_address_id and forwarding_date (ISO YYYY-MM-DD) are required.
Webhook events: mailing_received, scan_requested, mailing_scanned, mailing_archived, mailing_trashed, mailing_shredded, mailing_added_to_shipment, mailing_removed_from_shipment, shipment_sent, package_tracking_updated.
Requirements
- Python >= 3.10
- A Dropscan personal access token
Installation
From a local clone:
python3 -m venv .venv
./.venv/bin/pip install .
Configuration
The token is read from the DROPSCAN_TOKEN environment variable and sent as Authorization: Bearer <token>. It is never hardcoded.
Claude Code
claude mcp add dropscan -e DROPSCAN_TOKEN="dspat-..." -- dropscan-mcp
Claude Desktop / generic MCP client
{
"mcpServers": {
"dropscan": {
"command": "/absolute/path/to/dropscan-mcp/.venv/bin/python",
"args": ["-m", "dropscan_mcp.server"],
"env": {
"DROPSCAN_TOKEN": "dspat-..."
}
}
}
}
To enable the write/action tools, add "DROPSCAN_ALLOW_WRITES": "1" to env. Leave it out to keep the server strictly read-only.
The server runs over stdio.
Environment variables
| Variable | Description |
|---|---|
DROPSCAN_TOKEN | Required. Dropscan personal access token (Bearer). |
DROPSCAN_ALLOW_WRITES | Set to 1/true to enable write/action tools. Default: disabled. |
DROPSCAN_DOWNLOAD_DIR | Default directory for downloaded PDFs/ZIPs/images. Default: ~/Downloads. |
License
MIT, see LICENSE.
Author
Fabian Bitter ([email protected])
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.