metamcp logo
metamcp
★ 2.8k
vs
AgentBench logo
AgentBench
★ 3.8k

metamcp vs AgentBench

metamcp: MetaMCP is an MCP proxy designed to dynamically aggregate multiple MCP servers into a single unified server, applying custom middlewares. It serves as an infrastructure to host these composed MCP servers via a unified endpoint, compatible with any MCP client through SSE, Streamable HTTP, or OpenAPI.; AgentBench: AgentBench is a comprehensive benchmark for evaluating Large Language Models (LLMs) as agents across diverse environments, now featuring a function-calling version integrated with AgentRL. It provides a containerized setup for various tasks like OS interaction, database operations, and web shopping, enabling robust and reproducible agent evaluation.

01

TL;DR

metamcp logo
metamcpObservability · ★ 2.8k
When to choose:
  • You want one endpoint for many MCP servers, grouped into namespaces and served over SSE, Streamable HTTP or OpenAPI.
  • You need API key or OIDC authentication and registration controls for a team.
  • You want to remix tools with overrides and annotations and test them in a built-in inspector.
When to skip:
  • You want a single binary: it needs Docker and a Postgres database.
  • You need active maintenance: the README itself notes maintenance delays, and our data shows the last push on 2026-06-22.
  • You are counting on pluggable middleware or MCP tool search: the README marks them as coming soon.
AgentBench logo
AgentBenchObservability · ★ 3.8k

Systematically benchmark the performance of various LLM-based agents.

02

Side-by-Side Comparison

Field
metamcp logometamcp
AgentBench logoAgentBench
Category
Observability
Observability
Stars
★ 2.8k
★ 3.8k
License
MIT
Apache-2.0
Install Command
docker compose up -d
pip install -r requirements.txt
Updated
1w ago
9mo ago
Open Source
Yes
Yes
Tags
MCP, Proxy, Orchestration
LLM Evaluation, Agent Benchmarking, Function Calling
03

Features

metamcp logometamcp
01Dynamically aggregate MCP servers into namespaces and expose unified endpoints (SSE, Streamable HTTP, OpenAPI) with authentication.
02Apply pluggable middlewares at the namespace level to intercept and transform MCP requests/responses.
03Utilize an enhanced MCP inspector with saved server configurations for easy debugging of MetaMCP endpoints.
04Override tool names, titles, descriptions, and attach custom MCP annotations per namespace.
05Secure authentication with API keys, MCP OAuth, OpenID Connect (OIDC) provider support, and flexible registration controls.
AgentBench logoAgentBench
01Comprehensive LLM-as-Agent Evaluation across diverse environments.
02Function Calling integration for advanced agent interaction.
03Fully containerized deployment using Docker Compose for reproducibility.
04Multi-task and multi-turn interaction for realistic agent assessment.
05Extensible framework for adding new evaluation tasks.
04

Use Cases

metamcp logometamcp
↳Aggregating and hosting multiple MCP servers under unified, authenticated endpoints with namespaces.
↳Remixing MCP servers by selecting specific tools and applying custom middlewares for observability and security.
↳Debugging MetaMCP endpoints and server configurations using its built-in enhanced inspector.
AgentBench logoAgentBench
↳Systematically benchmark the performance of various LLM-based agents.
↳Develop and refine advanced LLM agent architectures and strategies.
↳Conduct academic research on the capabilities and limitations of agentic AI.
05

Best For

metamcp logometamcp
EssentialTrending
AgentBench logoAgentBench
Essential
FAQ

FAQ

What is the difference between metamcp and AgentBench?
Both metamcp and AgentBench are in the Observability category. metamcp has 2.8k stars, while AgentBench has 3.8k stars.
Which is better, metamcp or AgentBench?
The best choice depends on your use case. Choose metamcp if Aggregating and hosting multiple MCP servers under unified, authenticated endpoints with namespaces., and AgentBench if Systematically benchmark the performance of various LLM-based agents..
Is metamcp free or open source?
Yes, metamcp is open source on GitHub (MIT).
Is AgentBench free or open source?
Yes, AgentBench is open source on GitHub (Apache-2.0).
Cloud Infrastructure
Ad

Deploy on DigitalOcean

Host your MCP servers, Docker containers, and AI agents with $200 free credit over 60 days.

Claim $200 Free Credit
→

Related