AgentStack (agentstack.tech)

io.github.agentstacktechv1.0.0Updated Oct 6, 2026

Backend for AI agents: one batched tool over ~668 actions (hosting, storage, auth, payments).

VerifiedStreamable HTTPWeb executableCloud & InfrastructureDeveloper ToolsBusiness & Commerce

Overview

AI-generated overview

AgentStack is a remote MCP backend that exposes one batched tool covering hundreds of actions for hosting, storage, auth, and payments.

What it does
AgentStack presents itself as a backend for AI agents, offering a single batched tool that spans roughly 668 catalog actions (the README cites 656). The documented action areas include hosting, storage, authentication, and payments, alongside RAG, sandboxes, subscriptions, and REST/OpenAPI access. It is reached as a remote streamable HTTP endpoint rather than a local package.
When to use it
Consider it when an assistant needs to provision or manage backend resources such as hosting, storage, auth, or payments through one MCP tool instead of many separate integrations. It suits builders who want agent-driven access to a hosted platform and its REST/SDK surface.
Requirements
A remote streamable HTTP connection to the AgentStack endpoint; no local runtime or package is declared. The manifest declares no authentication, environment variables, or headers, though the documentation mentions OAuth 2.1 device code in a plugin and an npm SDK for building products.
Before you install
The manifest declares no authentication, so verify how access is actually controlled before connecting. The action set covers payments and authentication, which can move money or change account access; treat write, send, and delete actions as consequential. The documentation also references RAG and sandboxes, so consider what data leaves your environment.

Installation

In SourceWeft

  1. Open AgentStack (agentstack.tech) in the dashboard and add it to a workspace.
  2. 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.

Other MCP clients

Add this to your client's mcpServers config.

{
  "mcpServers": {
    "agentstack": {
      "type": "http",
      "url": "https://agentstack.tech/mcp"
    }
  }
}

README

AgentStack — Documentation

[platform] [MCP actions]

This repository contains the public documentation for AgentStack (English, user- and integrator-facing): web product, MCP, plugins, REST APIs, RAG, sandboxes, subscriptions, and examples.

If you use the website (dashboard): start with docs/USER_FEATURES_GUIDE.md — RAG, sandboxes, field access, subscriptions, dual-shell surfaces, and dashboard modules in plain language.

If you build your own product (app, site, client): docs/BUILD_YOUR_PRODUCT.md — golden paths, @agentstack/sdk, optional genetic-ai-starter.

If you run AI agents on a large codebase: docs/genetic-system/ — Navigation OS overview, economics (labor vs tokens), link to the portable kit.

If you integrate AI or automate: docs/MCP_AND_ECOSYSTEM.md — MCP (656 catalog actions — docs/MCP_SCALE.md), REST, plugins, SDK.

REST / OpenAPI: docs/OPENAPI.md — Swagger UI · openapi.json · API overview.

npm SDK: @agentstack/sdk · docs/sdk/

Full index: docs/README.md · What's new: docs/WHATS_NEW.md

Open source kit (new projects)

RepoBranchNotes
genetic-ai-startermainMap-first install: npx @agentstack/genetic-ai-starter init — philosophy, AI_NAVIGATION_MAP, Cursor rules. Docs: GENETIC_SYSTEM_ECONOMICS. SoT lives in AgentStack /genetic-ai-starter/.

Plugins (separate repositories)

PluginNotes
cursor-pluginv0.4.18, gen3 — rules, skills, commands, agents, hooks; Plugin MCP Connect and OAuth 2.1 device code
claude-pluginClaude Desktop / API installers
gpt-pluginChatGPT / OpenAI ecosystem
vscode-pluginVS Code marketplace distribution

Product

Source: README.md at commit 00ed350

Tools

0
Tool metadata has not been indexed yet.

Version history

1
  1. v1.0.0LatestOct 6, 2026