You don't need to read complete API endpoint documentation to understand how to configure Litmus products. You just need to ask the right question — and point your AI of choice at api.litmus.io
What does that mean in practice? api.litmus.io is Litmus's unified API portal — a machine-readable, always-current reference for the full REST and GraphQL of Litmus Edge, Litmus Edge Manager, and Litmus Unify. It's built not just for developers, but explicitly for AI: it ships a structured /llms.txt endpoint and an /agents context page that any LLM can fetch at runtime to get accurate, current knowledge of what the Litmus API can do.
Whether you use Claude, GPT-4o, Gemini, Copilot, or any other LLM — if it can fetch a URL, it can read the Litmus API docs and turn your natural-language questions into production-ready configuration.
This is not about AI generating guesses based on training data. Every time an AI fetches api.litmus.io, it reads the live documentation — so the configuration it produces is accurate, not approximate.
The industrial data space has always had a gap between knowing what you want to build and knowing exactly how to configure the system to build it. For OT engineers, that gap is API schemas and GraphQL mutations. For IT architects, it's translating operational requirements into repeatable pipeline definitions.
With api.litmus.io as the AI grounding layer, that gap closes. Here's what teams are already doing:
Analytics pipeline design: Ask an LLM to design an OEE, uptime, or anomaly detection pipeline and get back a step-by-step configuration guide with the exact API calls to build it.
Device and tag onboarding: Describe a new machine type and get the DeviceHub configuration to connect it, including driver settings and tag hierarchies.
Data model scaffolding: Define a machine template in plain language; the AI outputs the Data Modelling mutations to create reusable asset structures across your sites.
Integration routing: Ask how to publish processed data to InfluxDB, an MQTT broker, or a cloud endpoint — and get the exact publish parameters back.
None of this requires the AI to have been trained on Litmus. It just needs access to api.litmus.io at query time.
The video below shows one example of what this looks like in practice. A user asks their AI assistant — with api.litmus.io as the reference — a single question:
using api.litmus.io, explain me how would I leverage Analytics to build a pipeline that shows me OEE
No other context is provided. No manual doc-reading. No prior Litmus knowledge required from the AI.
Watch what happens after the prompt is submitted. The AI fetches api.litmus.io directly — you can see the live requests in the response trace. It reads the Analytics API reference, understands the pipeline model, and returns a complete implementation guide. No hallucination. No generic advice.
The response covers the full OEE pipeline architecture:
SOURCE layer: Three DataHub Subscribe processors subscribing to machineState, goodCount, and scrapCount topics from DeviceHub.
TRANSFORM layer: Production Time/CTR (Availability KPI), Compliance and Loss (Quality KPI), and Moving Average (Performance KPI with window_size: 60) — each as a separate, purpose-built processor. OEE combiner: A JSONata processor computing A × P × Q with a live timestamp output.
SINK layer: Database Output to InfluxDB (measurement: OEEData) and DataHub Publish to analytics.publish.* for downstream subscribers.
This is one prompt. The same approach works for any Litmus product configuration task — device onboarding, data model creation, integration setup, application deployment.
api.litmus.io covers over 1,900 REST and GraphQL endpoints across Litmus Edge, Litmus Edge Manager, and Litmus Unify — including DeviceHub, DataHub, Digital Twins, Analytics, Applications, OPC UA, and edge lifecycle management.
Two features make it purpose-built for AI consumption:
/llms.txt — A structured, machine-readable overview of the entire API surface. Any LLM that fetches this file gets an accurate map of what endpoints exist and what they do — without having to crawl individual pages.
/agents — A workflow context page that describes how Litmus operations map to API call sequences. Where the raw reference tells an AI what exists, this page explains how common tasks are actually accomplished — the ordered steps, the call dependencies, the expected returns.
The portal also includes end-to-end workflow examples. These show the exact sequence of API calls behind common Litmus Edge operations — the same kind of context a Litmus expert would give you in a design session, now accessible by any AI tool at runtime.
This architecture means api.litmus.io works with any MCP-compatible AI tool: Claude, GitHub Copilot, Cursor, Windsurf, or any custom agent that can fetch a URL and reason about structured documentation.
The full API reference, workflow examples, and AI agent endpoints are available at api.litmus.io. Point any LLM at it and start asking questions about the Litmus configuration task you're working on.
For deeper documentation on specific Litmus capabilities — Analytics, DeviceHub, Data Modelling, or edge application deployment — visit docs.litmus.io.
- 1.
Does this only work with one specific AI assistant? No. api.litmus.io is LLM-agnostic. Any AI tool that can fetch a URL — Claude, GPT-4o, Gemini, Copilot, Cursor, or a custom agent — can read the Litmus API reference and produce accurate configuration guidance. The /llms.txt and /agents endpoints follow open web standards, not a proprietary protocol.
- 2.
Is the AI reading live documentation or cached training data? Live documentation. When an AI fetch api.litmus.io at query time, it reads the current API reference — not a snapshot from training. This is what ensures the configuration it produces is accurate for your version of Litmus product, not an approximation from months-old data.
- 3.
What Litmus products does api.litmus.io cover? The portal covers Litmus Edge, Litmus Edge Manager, and Litmus Unify — all REST and GraphQL APIs across DeviceHub, DataHub, Digital Twins, Flows Manager, Analytics, Applications, OPC UA, UNS pub/sub, and edge lifecycle management.
- 4.
What's the fastest way to try this? Open your AI assistant of choice, paste a prompt that references api.litmus.io, and describe what you want to configure. For example: " Using api.litmus.io, help me configure the devicehub for integrating Fanuc CNC machine". The AI will fetch the docs and walk you through step by step.