Edgaze

ai.edgazePublished by edgaze.aiOfficial publisherv1.2.3Updated Oct 24, 2026

Find and run AI workflows from the Edgaze marketplace in your agent. Sign in with OAuth.

VerifiedStreamable HTTPWeb executableProductivity & WorkflowAI & ML

Overview

AI-generated overview

Remote MCP server for the Edgaze marketplace that lets an assistant search published AI workflows, inspect their inputs and price, and run them.

What you can do
  • Browse workflow categories when the goal is still vague
  • Search the marketplace for a workflow that fits a job
  • Inspect a workflow's inputs, example values, price and demo availability
  • Start a free demo or a paid run, with a cost ceiling set by the agent
  • Read run output until it finishes; text and external actions return in chat, media as a hosted run link
When to use it
Useful when you want an assistant to find and execute ready-made AI workflows from a marketplace instead of building them yourself, for example drafting outreach emails or other repeatable tasks, and you are willing to sign in to Edgaze and possibly fund a wallet for paid runs.
Requirements
A remote HTTP MCP endpoint at mcp.edgaze.ai; no local package, command, API key or Authorization header. Browser OAuth sign-in to Edgaze with consent for finding and running workflows. An Edgaze wallet with funds is needed only for paid runs; demos are free.
Before you install
Paid runs are charged to your Edgaze wallet, capped by the max_cost_usd ceiling the agent sets, so check the price before running. Your search terms, chosen workflow and inputs are sent to Edgaze, which runs the workflow on its servers. Some workflows act on your own connected accounts such as Slack or GitHub; inspect lists these before a run. Your client stores the OAuth token.

Installation

In SourceWeft

  1. Open Edgaze in the dashboard and add it to a workspace.
  2. Connect the required credentials. They are stored privately and never shown on public pages.
  3. 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/ai.edgaze%2Fedgaze/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.

Read the install guide the agent follows

Other MCP clients

Add this to your client's mcpServers config. The client completes OAuth sign-in on first connect.

{
  "mcpServers": {
    "edgaze": {
      "type": "http",
      "url": "https://mcp.edgaze.ai/"
    }
  }
}

README

Edgaze MCP

[CI]

Remote MCP server for the Edgaze marketplace. An agent can search published workflows, read the price and inputs, then run them.

This repository is public manifests only. The server is hosted. There is no local binary, no npx package, and no API key.

Server: https://mcp.edgaze.ai

Docs: https://www.edgaze.ai/docs/mcp

Connect

Every client connects the same way: add a remote HTTP server named edgaze at https://mcp.edgaze.ai. The client starts OAuth and opens consent on www.edgaze.ai. Approve Find workflows and Run workflows. Do not add a command, a bearer token, or an Authorization header, and do not open /oauth/consent yourself.

VS Code and GitHub Copilot

Workspace file .vscode/mcp.json (or Command Palette → MCP: Open User Configuration for a personal one). The top-level key is servers, and a remote server needs "type": "http":

json
{  "servers": {    "edgaze": {      "type": "http",      "url": "https://mcp.edgaze.ai"    }  }}

Or run MCP: Add Server, choose HTTP, and enter https://mcp.edgaze.ai. Confirm that you trust the server, then sign in to Edgaze. Switch Copilot Chat to Agent mode; Configure Tools shows the Edgaze tools.

Claude Code

Install the plugin. It adds the MCP server and a skill that tells Claude how to search, inspect, run, and show results.

text
/plugin install edgaze --marketplace edgaze-ai/mcp

From a shell:

bash
claude plugin install edgaze --marketplace edgaze-ai/mcp

Run /mcp, pick edgaze, and sign in when the browser opens.

Cursor, Claude app, Codex, and others

Snippets for each client: docs/clients.md. The full playbook with troubleshooting is Connect Edgaze MCP.

Tools

After you connect, these tools are listed:

ToolRole
edgaze_catalogCategories when the goal is still vague
edgaze_searchFind a workflow for a job
edgaze_inspectInputs, example values, price, demo availability
edgaze_runStart a demo or a paid run
edgaze_watchRead output until the run finishes

A zero wallet is normal the first time. Add money at https://www.edgaze.ai/wallet when you want a paid run.

Example prompts

  • "What kinds of workflows can I run on Edgaze?"
  • "Find an Edgaze workflow that writes a cold outreach email, and show me its price and the inputs it needs."
  • "Run a free demo of that workflow for my company, Acme Analytics, if it has one."
  • "Run it for real with a $2 ceiling and give me the email it writes."

What it does with your data

The server is hosted. The Claude plugin in this repository contains one MCP server declaration and one skill; it runs no local code, installs no packages, and adds no hooks.

  • Connects to: https://mcp.edgaze.ai only, over HTTPS.
  • Sends: your search terms, the workflow you pick, and the inputs you give it. Edgaze runs the workflow on its servers.
  • Signs in: with browser OAuth on www.edgaze.ai. Your client stores the OAuth token. Nothing in this repository reads or stores credentials.
  • Spends: a paid run is charged to your Edgaze wallet, never above the max_cost_usd ceiling the agent sets. Demos never bill. The skill tells the agent to read the price first and hide nothing about it.
  • Acts elsewhere: some workflows act on your own connected accounts, such as Slack or GitHub. edgaze_inspect lists these before you run.

Troubleshooting

  • No Edgaze tools in VS Code. The config needs the top-level servers key and "type": "http". Check that the server is started and trusted under MCP: List Servers, and that Copilot Chat is in Agent mode.
  • No Edgaze tools in Claude Code. Run /mcp and sign in to edgaze. A server you add by hand needs "type": "http", or Claude Code skips it.
  • No Edgaze tools on claude.ai or Claude Desktop. The connector is missing or off. Add it under Customize → Connectors → + Add → Add custom connector with the URL https://mcp.edgaze.ai. If it is already added, open + → Connectors in the chat and turn it on.
  • Consent says it is missing a sign-in request. The consent page was opened by hand. Use the client's Connect or Sign in control instead.
  • A run asks for money. The install worked. Add funds at https://www.edgaze.ai/wallet and run again.

More fixes: Connect Edgaze MCP.

After a run

Text and external actions come back in the chat. Images, video, and files come back as a link to the hosted run page: https://www.edgaze.ai/run/{id}.

Privacy and security

Sign-in is browser OAuth against Edgaze. The plugin stores no credentials.

Report vulnerabilities to [email protected]. See SECURITY.md.

Repository

FilePurpose
server.jsonOfficial MCP registry entry (ai.edgaze/edgaze), published from this repository
.claude-plugin/ / .mcp.jsonClaude Code plugin, marketplace, and directory listing
.cursor-plugin/ / mcp.jsonCursor plugin
skills/edgaze/SKILL.mdHow the agent should search, inspect, run, and show results

Implementation is not accepted here. Changes to the hosted server happen in the Edgaze product.

License

MIT. Copyright Edge Platforms, Inc.

Source: README.md at commit 2c24daf

Tools

0
This server asks you to sign in, so SourceWeft could not read its tools without an account and has not indexed them. What a connected account sees can depend on that account and its permissions.

Version history

2
  1. v1.2.3LatestOct 15, 2026
  2. v1.0.0Oct 12, 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