Recently updated agents and stale candidates

Metorial lets developers connect any AI model to over 1200 external integrations through MCP, a CLI, or an API. It acts as a universal bridge between AI runtimes and services like databases, file storage, or messaging platforms.

A collection of Python adapters that let agent frameworks like LangChain, LlamaIndex, CrewAI, and n8n submit results to the P2PCLAW leaderboard. Each adapter handles the protocol translation so agents can benchmark against a shared scoring system.

Cellar is a MCP server that provides structured perception and reliable actions for native computer use on macOS. It enables continuous awareness of the system state through a computer experience layer.

A Python library that bridges meeting platforms with AI agent workflows. It enables AI agents to join, listen, and interact in virtual meetings by providing a unified interface for real-time audio and video streams.

BabyBeeAGI is a lightweight modification of BabyAGI with about 300 lines of code. It automates task generation and execution using GPT-4, but it is slower and buggier than the original. Builders can fork the Replit project, add an API key, and set an objective to experiment with autonomous agent workflows.

A Python library for graph-based tool retrieval in LLM agents. It selects relevant tools from a set of up to 248 tools using a graph structure, achieving 82% accuracy while consuming 79% fewer tokens than baseline approaches. The library has zero external dependencies and supports OpenAPI, MCP, and LangChain integrations.

Open‑source SDK for building and orchestrating single or multi‑agent systems with LLMs and tool integration.

Markdown to PDF conversion, for agents.

Airtable integration for AI-powered applications via Anthropic's Model Context Protocol (MCP). Connect your AI tools directly to Airtable for seamless data access and management.