
Vepathos
com.vepathosPublished by vepathos.comOfficial publisherv0.11.1Updated Oct 24, 2026
Import orders, manage your fleet, and generate optimized last-mile delivery routes at scale.
Overview
Vepathos is a remote MCP server that lets an assistant import deliveries, manage fleet master data and generate optimized last-mile delivery routes.
- What you can do
- Check the connected account, its plan and available plans, and list saved fleet, depots, vehicles and automations, plus past imports and their results. Import deliveries and correct an import's column mapping without re-uploading. Geocode street addresses into coordinates and poll the job status. Plan and optimize vehicle routes from a depot, then fetch the status and outcome of an optimization by its id. Add or change vehicles, depots and fleets saved in the account, and prepare scheduled routing rules that stay switched off. Create a temporary public link to the map of a completed optimization.
- When to use it
- Use it when an assistant should plan last-mile delivery routes for a fleet, from dozens to thousands of stops, or keep fleet and depot master data in a Vepathos account. It suits logistics and operations work where addresses must be geocoded and routes optimized against vehicle capacity and time windows.
- Requirements
- A remote endpoint at mcp.vepathos.com over streamable HTTP, with OAuth sign-in or sign-up and authorization of the client against a Vepathos account. Headless use needs a dashboard credential with scope mcp:optimize sent as an Authorization Bearer header. Usage counts against the account's plan and monthly stop quota.
Installation
In SourceWeft
- Open Vepathos in the dashboard and add it to a workspace.
- Connect the required credentials. They are stored privately and never shown on public pages.
- Enable the server for the chats that should use its tools.
Web executable via Streamable HTTP. Remote servers run from the web runtime once configured in a workspace.
Use with an agent
Paste this prompt into Claude Code, Codex, Cursor or another agent — or into SourceWeft chat. The agent reads this server's install guide, shows you its publisher, endpoint and sign-in, and adds it to your client once you agree. Keys never go into a command.
Read https://sourceweft.com/mcp/com.vepathos%2Fvepathos/install.md and connect the MCP server it describes to my client. Before changing any config, show me its publisher, endpoint and sign-in method and wait for my OK. Never put an API key in a command or in a file I would commit; ask me where it should come from.Other MCP clients
Add this to your client's mcpServers config. The client completes OAuth sign-in on first connect.
{
"mcpServers": {
"vepathos": {
"type": "http",
"url": "https://mcp.vepathos.com/mcp"
}
}
}README
Vepathos MCP
Large-scale delivery and fleet optimization for AI agents.
Vepathos solves vehicle routing problems (VRP) for last-mile fleets: it assigns stops to vehicles and sequences each route from one depot, at scales from dozens to thousands of stops.
This repository is the remote MCP adapter (mcp.vepathos.com). It is not a product of its own.
Web, REST and MCP share the same Vepathos account, plan, features, limits and monthly stop quota.
No tool switches an automation on: create_automation always writes it switched off, and only its
owner turns it on in the dashboard, because a rule that is on spends their stops unattended.
There is no cancel tool. A submitted optimization runs to completion. Street addresses must go
through geocode_addresses (or, with imports on, import_deliveries) first; optimize_routes does not
invent coordinates.
Status
Production runs 0.11.1 at https://mcp.vepathos.com/mcp and publishes all 15 tools: the
import tools, the shareable map and manage_catalog are switched on there, so the "Available"
column above describes what the code gates, not what this deployment withholds.
manage_catalog saves three kinds of thing: a vehicle, a depot, and a fleet grouping vehicles the
account already has. A fleet's units is what it holds standing, never how many go out on one run,
which is vehicles[].count on optimize_routes. To run with a saved fleet, read it with
list_fleet and pass its vehicles: optimize_routes takes no fleet_id.
Vepathos is listed in the official MCP registry as
com.vepathos/vepathos, on Smithery, and on
Glama. Directories take a snapshot when they
crawl, so a listing can lag a deployment by a day; the registry entry is republished on every
version bump.
Paid self-serve is off. A request the account's plan cannot run answers PLAN_UPGRADE_REQUIRED
with a contact_url, never a Stripe Checkout link, and the user retries without reconnecting once
the account can run it.
Two ways to run it away from production:
- Local and CI: this server against a fake Core, a test double that neither routes nor
geocodes for real. Everything in
tests/uses it. - Local against the real stack: the
vepathos-api-docMCP channel, the optimizer and the Smart Import worker.
Connect (production target)
- Discover Vepathos from Claude or another MCP client.
- Connect. The client signs in (or creates a Free / Duck account) at
api.vepathos.com. - Authorize the client to optimize routes with that account.
- Call
optimize_routes. If the plan cannot run the request, the tool returnsPLAN_UPGRADE_REQUIREDwithupgrade_url(when paid plans are on) orcontact_url(Free-only, until Stripe is configured). Payment, when enabled, is handled entirely by Stripe. Retry without reconnecting after the account can run the job.
No API keys and no JSON config for that flow. See docs/onboarding.md.
Developers and headless agents
Create a Vepathos dashboard credential with scope mcp:optimize and send
Authorization: Bearer <client_id>:<client_secret>
to https://mcp.vepathos.com/mcp. Usage counts against the same account plan as the web app and
the REST API.
Example (3,200 deliveries)
Every stop needs latitude and longitude. Addresses are not geocoded. Weight, volume and time
windows are optional; if you set a capacity on any vehicle, every vehicle and every stop must
include that field. Time windows require schedule.route_start_time.
Results stay available for 24 hours. detail=summary is compact (totals and a page of routes).
detail=stops returns ordered stop_id + arrival time, without echoing coordinates.
Full schemas and errors: docs/tools.md.
Run locally
Requires Python 3.12+.
The MCP endpoint is http://127.0.0.1:8080/mcp with
Authorization: Bearer dev-bearer-token-change-me. That compose stack talks to the fake Core, not
the optimizer.
MCP Inspector and the integration against the real local stack: docs/testing.md.
Privacy
The tools accept coordinates, optional weight/volume/time windows and your own stop and vehicle
ids. Street addresses reach Smart Import through geocode_addresses and import_deliveries and are
not echoed back. Results do not repeat coordinates, logs omit tokens, payloads and coordinates, and
hosted results are kept for 24 hours.
Account and logistics data follow the public policy at vepathos.com/privacy, which carries an MCP section. See also docs/security.md and SECURITY.md.
Research & benchmarks
The Vepathos last-mile optimizer is described in a public technical report: doi:10.5281/zenodo.19859531. That deposit is not a peer-reviewed publication.
Docs
License
Source: README.md at commit 1d74cee
Tools
0Version history
2- v0.11.1LatestSep 25, 2026
- v0.10.0Sep 24, 2026
Score
85% of the applicable checks met
How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.
- Reached in the last 30 daysRequired25 pts
- Auth declaredRequired10 pts
- Tools listed15 pts
Answer tools/list without a credential, or ship a package that starts in a sandbox without one.
- DescriptionRequired15 pts
- Logo10 pts
- Documentation15 pts
- License10 pts