Log inSubmit server
Wiro MCP logo

Wiro MCP

Run Wiro AI models, workflows, and agents through MCP from compatible assistants.

InstallRepositoryWebsite
Full README Compare alternatives Embed widget Claim this listing

Install Wiro MCP

Pick your client, then add it in one click or paste the config.

STDIONode.js
~/Library/Application Support/Claude/claude_desktop_config.json
claude_desktop_config.json
{
  "mcpServers": {
    "wiro-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "@wiro-ai/wiro-mcp"
      ],
      "env": {
        "WIRO_API_KEY": "YOUR_VALUE_HERE",
        "WIRO_API_SECRET": "YOUR_VALUE_HERE"
      }
    }
  }
}

Settings → Developer → Edit Config, merge this into mcpServers, then fully quit and reopen Claude Desktop.

Values are written into the snippet above so you can paste it as is. They never leave your browser.

Install not yet sandbox-tested— we haven't run this command through our automated sandbox yet. Not a red flag; we're working through the catalog.

Overview

Wiro-MCP MCP server connects MCP-compatible assistants to Wiro AI's catalog of image, video, audio, and LLM models. It supports model discovery, parameter and pricing lookup, asynchronous task handling, file uploads, workflow creation and execution, and conversations with Studio or deployed agents. The local package runs over stdio through npx, while Wiro also provides a hosted streamable-HTTP endpoint. Reach for it when an assistant needs to generate media, compose multi-step model workflows, or interact with Wiro agents.

Use cases

  • Generate images, videos, audio, or text with Wiro models
  • Compare model parameters and prices before running a task
  • Build and test multi-step media workflows
  • Resume failed workflow runs without repeating completed steps
  • Message Studio or manage conversations with Wiro agents

Key features

  • Wiro model catalog search and recommendations
  • Asynchronous task execution and recovery
  • Public URL file uploads
  • Workflow authoring, execution, and resumption
  • Studio and agent conversations
  • Agent template and result browsing

Tools50

What Wiro MCP lets your agent do. Expand a tool for its parameters.

Self-reported~1.2k tokens of context

Security & health

What it can reach and how reliably it runs — from our own checks, not the maintainer’s claims.

ActiveLast checked Today

Recent health check succeeded.

Supply-chain signal

No high-severity advisories surfaced by our automated scan.

Critical 0High 0Medium 0Low 0

Scanned 8d ago via OSV.dev · @wiro-ai/wiro-mcp (npm)

Adoption & maintenance

Factual signals from GitHub, npm, and our automated checks — not a rating.

GitHub stars4Stargazers on the source repository.
npm downloads2.2kPackage downloads in the last 30 days.
Last commit2d agoMost recent push to the default branch.
Tools exposed50Callable tools this server registers over MCP.

How Wiro MCP works

What Wiro-MCP does

Wiro-MCP exposes Wiro AI capabilities as Model Context Protocol tools. Assistants can search the model catalog, compare models for a stated goal, inspect input schemas and pricing, and run models for image, video, audio, or language tasks. The server also provides access to task history, result pricing, cancellation, and file upload from a public HTTP(S) URL.

Its second major area is workflow automation. A workflow combines model and utility steps into a graph with defined inputs and outputs. Agents can browse a workspace or public gallery, inspect existing graphs, create or update workflows, clone public workflows, execute them, and resume eligible failed or stopped runs. Workflow Agent proposals can also be saved after confirmation.

The Agents tools support conversations with Wiro's Studio creative agent and access to agent templates. They can send messages, retrieve replies and conversation history, list Studio results, and manage agent sessions.

How it works

The npm package is a TypeScript MCP server that runs locally over stdio. It can be started with npx, allowing an MCP client to launch it as a subprocess. Wiro also hosts the same tool set at https://mcp.wiro.ai/v1 over streamable HTTP, which avoids installing Node.js locally.

Model and workflow execution can take longer than a single tool call. run_model waits for a result for up to 45 seconds by default, then returns a task identifier and recovery token if the work is still running. Continuing with wait_for_task avoids submitting and billing a duplicate run. Workflow runs have corresponding wait and resume operations, including the ability to continue after a failure without paying again for completed steps.

Workflows are checked against Wiro's live model catalog when saved. Missing model inputs use the model's defaults, and the save response reports those filled values. Workflow graphs are limited to 40 KB of JSON.

Setup and configuration

Create a Wiro AI account and project to obtain credentials. Local execution requires Node.js 20 or later. Configure the MCP client to launch npx -y @wiro-ai/wiro-mcp and pass the credentials as environment variables.

Signature-based authentication uses WIRO_API_KEY and WIRO_API_SECRET. API-key-only authentication uses WIRO_API_KEY alone, with the secret omitted. For the hosted endpoint, send the credentials as a bearer token: API_KEY:API_SECRET for signature-based projects or just API_KEY for API-key-only projects.

The server includes documented configurations for Cursor, Claude Code, Claude Desktop, and Windsurf. Remote clients can connect to the hosted endpoint instead of running the local package.

Tools and capabilities

  • Discover models with keyword, category, owner, curated-category, or goal-based searches.
  • Read model parameters, output ports, and pricing before running a model or utility.
  • Execute models, wait for tasks, inspect task status and history, retrieve completed-task prices, and cancel or stop work.
  • Upload public HTTP(S) files of about 100 MB or less as model inputs.
  • List, inspect, create, edit, clone, delete, run, and resume workflows.
  • Test a workflow through a selected step and inspect failed-step details.
  • Save Workflow Agent proposals after reviewing the proposed change.
  • Message Studio or other agents, retrieve replies, inspect conversations, cancel replies, delete sessions, and browse generated Studio results.
  • Browse agent templates and their monthly prices.

Limitations and notes

Model runs and completed workflow model steps can incur charges. Retrying a model run after its wait period creates a separate billable task, so existing task identifiers should be continued with the wait tool instead. Workflow deletion and other consequential changes require confirmation through the relevant tool flow. Public-file uploads require an accessible HTTP(S) URL, and workflow graphs cannot exceed 40 KB of JSON.

Read the full README View source on GitHub

Reviews

No reviews yet — be the first to share how it worked for you.

Frequently Asked Questions about Wiro MCP

Wiro-MCP is an MCP server that connects compatible AI assistants to Wiro AI's models, workflows, Studio, and agents. Its main tools search models, inspect schemas and prices, run and monitor tasks, upload inputs, create and execute workflows, and send or retrieve agent messages through local stdio or a hosted HTTP endpoint.

Run it locally with Node.js 20 or later using `npx -y @wiro-ai/wiro-mcp`. Configure your MCP client to launch that command and provide `WIRO_API_KEY`; add `WIRO_API_SECRET` when using signature-based authentication. You can instead connect to `https://mcp.wiro.ai/v1` without a local Node.js installation.

Yes, Wiro-MCP works with Cursor. Add an MCP server entry that runs `npx -y @wiro-ai/wiro-mcp`, then set `WIRO_API_KEY` and, for signature-based projects, `WIRO_API_SECRET` in the entry's environment.

Continue the existing task with `wait_for_task` instead of calling `run_model` again. When its wait period expires, `run_model` returns the existing task identifier and a recovery token, while submitting another run creates a separate billable task.

Yes, `resume_workflow_run` continues a failed or stopped workflow without charging again for steps that already finished. Use `wait_for_workflow_run` to keep waiting, read outputs, or inspect the step that failed.

Related MCP servers

All in Conversational AI All alternatives
Add a badge to your READMEShow it’s listed — live SVG, Markdown or HTML.

AllMCPs Directory Badge

Full Badge Customizer

Showcase your server listing on GitHub or your project documentation. Embed this dynamic SVG badge to highlight official listing status and live engagement.

Badge Style:
Live Dynamic SVG PreviewWiro MCP AllMCPs Directory Badge
Markdown (GitHub README)
[![AllMCPs](https://allmcps.com/api/badge/wiro-mcp?style=directory)](https://allmcps.com/mcp/wiro-mcp)
HTML Embed
<a href="https://allmcps.com/mcp/wiro-mcp"><img src="https://allmcps.com/api/badge/wiro-mcp?style=directory" alt="Wiro MCP on AllMCPs" /></a>