# AgentoolRank [Health: Active]

**Category:** 💻 Developer Tools  
**Repository:** https://github.com/agent-gigmole/agentoolrank-mcp  
**GitHub Stars:** 0  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/agentoolrank

## Description
Search and compare open-source AI agent tools ranked by live GitHub activity, with compare prompts.

## Claude Desktop Quick Installation
Remote MCP endpoint (confidence: high). Install path detected from listing signals. Add as a URL/SSE server in your client:

```json
"mcpServers": {
  "agentoolrank": {
    "url": "https://agentoolrank.com/api/mcp"
  }
}
```

## Documentation

## What AgentoolRank MCP server does

AgentoolRank MCP server exposes a hosted catalog of open-source AI agent tools through Streamable HTTP at `https://agentoolrank.com/api/mcp`. The catalog covers categories such as frameworks, coding agents, memory, RAG, evaluations, and MCP servers. Rankings are refreshed daily using GitHub stars, 30-day star growth, commits, releases, and package downloads.

The service supports both research and product-listing workflows. An agent can search for tools, inspect a tool's capabilities and limitations, find open-source alternatives, or compare two tools through a ready-made prompt. It can also submit a tool, check a submission's status, recommend launch directories, and manage the Submit Kit workflow.

## How it works

MCP clients connect to the hosted endpoint over Streamable HTTP; there is no local installation or self-hosting step. The server does not require authentication for its MCP endpoint. An optional API key can be obtained from AgentoolRank's API-key page and sent as an `Authorization: Bearer <key>` header to identify the agent.

The available tools include `search_tools`, `get_tool`, and `get_alternatives` for discovery. Submission-related tools include `submit_tool` and `get_submission_status`. Directory workflow tools provide recommendations, the next directory to process, result reporting, and progress checks through `recommend_directories`, `next_directory`, `report_result`, and `my_progress`.

The server also publishes the `compare_tools`, `find_alternatives`, and `launch_plan` prompts. Its resources include `agentoolrank://about` and `agentoolrank://top-tools`, the latter providing a live table of the current top 20 tools. The same underlying data is available through JSON endpoints for clients that do not use MCP.

## Setup and configuration

AgentoolRank MCP server can be added directly to Claude Code with an HTTP transport command. Cursor uses an MCP configuration entry containing the service URL, while Claude Desktop can use a custom connector or the documented `mcp-remote` bridge. Codex CLI uses its TOML configuration. The repository includes configuration examples for these clients.

The JSON API provides a search endpoint at `/api/v1/tools`, a detail endpoint at `/api/v1/tools/{slug}`, and a POST endpoint at `/api/v1/submissions`. API-key use is optional, and the supplied material does not identify any required environment variables.

## Limitations and notes

The ranking reflects the service's tracked GitHub and package activity signals and is refreshed daily, so it should not be treated as a real-time or complete measure of tool quality. The repository's MIT license applies to its documentation and configuration examples; the AgentoolRank website, service, and data are excluded from that repository license.

Directory submission reports are leads, while listing rates count only listings verified by the service's daily check. The Submit Kit workflow limits unreported directories and hands them out one at a time. The hosted service is operated by TENSO LLC, and the provided material does not describe a local server package or self-hosting procedure.

_Full upstream README: https://allmcps.com/mcp/agentoolrank/readme_