01 · Docs
The MIP reference is not public
MPDV does not publish the MIP API reference or the SDK. Access runs through MPDV or a certified partner, so the tool list is written against your installation's services, not against a public spec.
For manufacturers running MPDV HYDRA X
A HYDRA X MCP server that reads orders, downtimes and scrap under each user's own HYDRA permissions
A HYDRA X MCP server gives Claude, ChatGPT or Copilot read tools for production orders, machine downtimes, scrap and shift figures. HYDRA X exposes MIP web services, but MPDV does not publish the API reference. MPDV ships no MCP server as of 21 September 2026, so Pexon builds one against your installation.
Why HYDRA X is hard for agents
Where a HYDRA X connection needs engineering, not configuration
02 · Versions
Service names differ by HYDRA generation
Services are addressed by MIP information-model type and service names. These differ between HYDRA 8 and HYDRA X installations, so each tool is mapped to the services your plant actually runs.
03 · Network
HYDRA sits next to the shop floor
Most installations run on premises, close to the machines. The MCP server has to run inside the plant network or reach it through a gateway, which is a network decision your IT team owns.
04 · Writes
Bookings touch running production
Order confirmations and bookings change what the shop floor sees in real time. Any write stays a proposal that an operator confirms in HYDRA.
Checked 21 September 2026
HYDRA X at a glance: interfaces, access and what exists for AI
| What we found | |
|---|---|
| Interfaces | Manufacturing Integration Platform (MIP) web services (REST/JSON, docs behind customer or partner login): HYDRA X applications communicate through MIP web services on a shared information model; no public API reference, access via MPDV or its partners · MIP SDK and mApps (REST/JSON, docs behind customer or partner login): customers and partners build their own mApps on MIP and offer them through the MIP Marketplace |
| Authentication | Not documented publicly by MPDV. A community Node-RED node for HYDRA X and HYDRA 8 calls the REST services with username, password and an AccessId; the method for your installation is confirmed in the blueprint. |
| Vendor MCP server | No vendor server. No MPDV MCP server found; the MIP and AI Suite pages do not mention MCP or AI agents. |
| Vendor AI assistant | The MPDV AI Suite for HYDRA X and FEDRA ships machine-learning mApps such as AI Planning, setup time prediction and predictive quality; it is analytics, not a conversational assistant. |
| Other connectors | innFactory builds a custom MCP server per customer on the MIP REST services for its CompanyGPT product, and a community Node-RED node calls the HYDRA REST API; neither is an MPDV product. |
MPDV Mikrolab GmbH (DE). Discrete and process manufacturers of all sizes, from automotive suppliers to food and beverage; MPDV reports 2,430 installations.
What we build
A first HYDRA X tool list, written as the manifest your IT team signs
Built on what MPDV Mikrolab GmbH documents: production orders, operations and work steps, machine states and downtime reasons, run times, scrap quantities, order confirmations, personnel and attendance, quality data. Backends named below are documented resources; "confirmed in blueprint" means the vendor documentation sits behind a login and the resource is checked against your release first.
# mpdv-hydra-mcp-server: tool manifest, draft for blueprint week 1
# Every tool runs as the person asking. Nothing is written without a person confirming.
tools:
get_order_status: # order number -> status, operations, good and scrap quantities, confirmations
backend: MIP order services, resource confirmed in blueprint
mode: read
list_machine_downtimes: # machine or group, time range -> downtime periods with reason codes
backend: confirmed in blueprint
mode: read
get_shift_summary: # workplace, shift -> run time, downtime, scrap, OEE figures
backend: confirmed in blueprint
mode: read
get_scrap_by_reason: # article or machine, date range -> scrap quantities grouped by reason
backend: confirmed in blueprint
mode: read
draft_downtime_reason: # downtime id, proposed reason -> draft for an operator to confirm
backend: confirmed in blueprint
mode: propose
audit: every call logged with user, tool, arguments and the HYDRA X responseThe manifest is the contract. A question that needs an operation not on it gets a new line and a review, not a wider permission.
HYDRA X holds the records. An assistant without a governed way in answers from whatever someone pasted into the chat. The connector is the governed way in: named operations, the person's own permissions, every call logged.
What the connector answers, and what it will not do
Questions it answers
- Which machines on line 3 had the most downtime in the last shift, and why?
- What is the status and remaining quantity of production order 4711?
- How did the scrap rate on machine group X develop over the last four weeks?
What it will not do
- Book, confirm or change anything by itself: the only write tool drafts a downtime reason and an operator confirms it in HYDRA.
- Read more than the asking user may read in HYDRA; permissions come from HYDRA and MIP, not from the connector.
- Replace the AI Suite: planning optimisation and predictive quality stay in MPDV's own mApps.
When this is the wrong page
If the goal is AI planning, setup time prediction or predictive quality inside HYDRA X, MPDV's AI Suite is the shorter path. A connector is right when people ask questions from Claude, ChatGPT or Copilot, or need HYDRA data next to ERP or quality data.
Start with the Readiness Blueprint.
Two weeks, €4,900 fixed price. Week one agrees the tool list with your HYDRA X owners and checks every backend against your release and licence. Week two tests whether a person's own HYDRA X permissions reach through the connector. You keep the manifest and a costed build plan, whoever builds it. All prices are net and exclude VAT.
What we need from you: a named HYDRA X administrator, a test or sandbox system we may call, and one business owner who can say which questions the assistant should answer first. The first production connector is usually a use-case pilot from €15,000; it runs in your environment, and the code is yours.
Related
- Connectors for 100 source systems — the dated catalogue this page belongs to
- Claude MCP server development — how we build an MCP server for any internal system
Questions HYDRA X teams ask before the first tool call
Does HYDRA X have its own MCP server?
Not as of 21 September 2026. MPDV's MIP and AI Suite pages do not mention MCP or AI agents, and the AI Suite is analytics inside HYDRA, not a conversational assistant. An MCP server for Claude, ChatGPT or Copilot has to be built against the MIP web services.
Do we need extra MPDV licences or access for an AI connector?
MPDV does not publish its API access terms, and the MIP reference is available through MPDV or a certified partner. Check with MPDV whether an external application on MIP needs its own licence. The blueprint settles this before any tool is built.
Can the assistant see data a user is not allowed to see in HYDRA?
No. Each call runs under the asking person's HYDRA identity, so HYDRA's own permission check decides every answer. Two users asking the same question can get two different answers, and that is tested before go-live.
What does a HYDRA X connector cost?
The two-week Readiness Blueprint is fixed at €4,900: agreed tool list, backends checked against your release, identity tested end to end, costed build plan. The production connector is usually a use-case pilot from €15,000. Prices are net, excluding VAT.
Next step
Not a sales call. An architecture call.
Thirty minutes with the engineer who would build the connector, with your HYDRA X administrator in the room if you like.