Runs a vision workflow on demand, validates a specification before it executes, and reports inference server health and metrics.
7actions available
Every action runs live through MCP. Nothing to build, nothing to maintain.
Ayro retrieves information about the Roboflow inference server. Operators monitor server health to catch issues early.
Ayro checks a Roboflow workflow specification for errors before it runs. Teams catch configuration mistakes instead of failing mid-processing.
Ayro executes a Roboflow workflow with your provided inputs. Vision AI pipelines process images automatically instead of manual analysis.
Each card is one request a support team gets, and the Roboflow actions Ayro runs to close it.
Ayro reads the workflow's declared inputs, outputs, and types, then executes the specification with the values supplied in the conversation and returns what the pipeline produced.
Ayro checks a workflow definition against the block schema and validates its structure, so a syntax problem surfaces as an answer rather than as a failed inference run.
Ayro reads the server's identity and version alongside its Prometheus metrics, so a question about slow or failing inference gets current numbers rather than a guess.
Ayro lists the Execution Engine versions the workflows API supports and reads the current block schema, so a compatibility question is settled before an upgrade is attempted.
How it works
Connect once, then just ask. There is no workflow builder to learn and nothing to maintain — Ayro reads the Roboflow actions it has and picks the ones a request needs.
Authorize the Roboflow account your team already uses — one consent screen, no API keys, no mapping tables. Ayro can only do what you granted that account, and you can disconnect it at any time.
Describe the job the way you would hand it to a teammate. Ayro maps it to the Roboflow actions that close it and chains as many as the request needs.
The agent runs the actions inside the conversation the customer is already in, so nobody copies data between tabs and your team can take over at any point.
Get started freeRoboflow · Ayro
What can you do in Roboflow without me setting anything up?
These, right away:
Great, let's start with the first one.
FAQ
Setup, permissions, and the limits of what Ayro can do inside Roboflow.
No. This toolkit covers workflow execution and inference-server inspection - Run Workflow, Validate Workflow, Get Server Info, Get Server Metrics. Training runs and dataset management stay in the Roboflow console.
All 7 actions your agent can call on Roboflow, straight from the live MCP connection.
Describe the interface of a workflow specification.
Retrieve available Execution Engine versions from Roboflow workflows API.
Retrieve information about the Roboflow inference server.
Retrieve Prometheus metrics from the Roboflow inference server.
Fetch the workflows block schema from Roboflow.
Run a workflow specification with provided inputs.
Validate a Roboflow workflow specification before execution.
Same connection, same setup. Pick the next one your team already uses.
Reads the available voice catalogue, returning each voice_id, name, gender, and language, and narrows it to one language code on request.
Searches the toolkit catalogue for the right action, walks users through authorising accounts, executes tools singly or in batches, and enables triggers.
Generates text, speech, images, and Veo video from a prompt, embeds text for semantic search, and counts tokens before a call is made.
Submits documents for conversion, tracks the job by its whisper hash, retrieves the extracted text, and manages the callback webhooks.
Runs a template against supplied data, reports app job status and results, and manages the webhook subscriptions behind those jobs.
Drafts product titles, social posts, emails, and video descriptions, then summarizes or translates them across 72+ languages.

Authorize the account and your agent has all 7 actions from the first conversation.
