AgentStackd MCP server
by piercewes77
README.md
# AgentStackd MCP server
A free, read-only MCP server for searching and comparing the 77 hand-reviewed AI agents and tools listed on [AgentStackd](https://agentstackd.com). No API key.
**Remote URL:** https://agentstackd.com/api/mcp
Streamable HTTP, MCP protocol `2025-06-18`. Listed on the official MCP Registry as `com.agentstackd/directory` v0.1.0.
Setup walkthrough: [agentstackd.com/mcp/](https://agentstackd.com/mcp/)
## Tools
**search_tools** — Search the directory by keyword, with optional category and pricing filters. Returns matching tools with pricing model, description and the agentstackd.com page URL.
Arguments:
- `query` (string, required) — Keywords, e.g. `"coding agent"`, `"web research"`, `"self-hosted automation"`.
- `category` (string, optional) — A category slug from `list_categories`. Current slugs: `coding`, `research`, `sales`, `support`, `content`, `assistants`, `automation`.
- `max_price` (string, optional) — Pricing ceiling: `free` (Free or Open source only), `freemium` (also tools with a free plan), or `paid` (everything). The directory records pricing models, not dollar amounts.
- `limit` (integer, optional) — Max results, 1–25 (default 10).
**get_tool** — Get one tool's listing: pricing model, description, pros, limitations, vendor sources (with the date they were checked), related comparisons and the agentstackd.com page URL.
Arguments:
- `slug` (string, required) — Tool slug (e.g. `"cursor"`, `"claude-code"`) or exact tool name.
**compare_tools** — Compare 2 or 3 tools side by side: category, pricing model, pros, limitations and sources, plus any AgentStackd head-to-head comparison page that covers them.
Arguments:
- `slugs` (array of 2 or 3 strings, required) — Tool slugs or names, e.g. `["claude-code", "openai-codex"]`.
**list_categories** — Categories with descriptions, tool counts and page URLs. No arguments.
**new_releases** — Recently released AI tools added to AgentStackd, with the vendor's release date and page URL. No arguments.
Every result includes the agentstackd.com page URL.
## Connect
### Claude Desktop
Custom connectors work on Claude and Claude Desktop for Free, Pro, Max, Ultra, Team and Enterprise plans (Free is limited to one). Go to Customize → Connectors, click + Add, then Add custom connector, and paste the endpoint URL:
```
https://agentstackd.com/api/mcp
```
Alternatively, add this to `claude_desktop_config.json` (needs Node.js for `mcp-remote`):
```json
{
"mcpServers": {
"agentstackd": {
"command": "npx",
"args": [
"-y",
"mcp-remote",
"https://agentstackd.com/api/mcp"
]
}
}
}
```
### Cursor
Add this to `~/.cursor/mcp.json` (or `.cursor/mcp.json` in a project):
```json
{
"mcpServers": {
"agentstackd": {
"url": "https://agentstackd.com/api/mcp"
}
}
}
```
### VS Code
Add this to `.mcp.json` at your workspace root, or to `~/.copilot/mcp-config.json` for all workspaces. You can also use Chat: Open Customizations → MCP Servers → Add Server, choose HTTP, and paste the endpoint URL:
```json
{
"mcpServers": {
"agentstackd": {
"type": "http",
"url": "https://agentstackd.com/api/mcp"
}
}
}
```
## Example prompts
Once connected, you can ask things like:
- Search AgentStackd for free coding agents.
- What's the pricing model and limitations of Cursor?
- Compare Claude Code and OpenAI Codex.
- What categories does AgentStackd cover?
- Which tools were added recently?
## Rate limits and access
Free, no API key, read-only. No login and no writes; we don't store your queries. Requests are rate limited to 60 per minute per IP.
## This repository
`api/mcp.js` is the live server (a Vercel Node function). It reads `api/_lib/data.json`, a snapshot of the public listing data generated when [agentstackd.com](https://agentstackd.com) is built. The live dataset may be newer than the snapshot in this repo.
`server.json` is the MCP Registry server description.
## License
MIT. See [LICENSE](LICENSE).
This server cannot be deployed
Maintenance
ActivityMaintained
ResponsivenessNo recent community issues