Open-Source

Orchestrates developer workflows, integrates with GitHub, and manages tasks locally via CLI with a persistent memory.

Automates computer tasks by allowing LLMs to control browsers, terminals, and file systems, dynamically learning and evolving skills over time.

Integrates Sentry error monitoring and performance tracing into AI coding assistants, enabling setup, issue identification, and configuration directly via natural language commands.

Provides a self-hosted web interface to interact with and manage the persistent, self-improving Hermes AI agent, enabling continuous learning and task automation.

Generates playable web games end-to-end from text prompts using an open-source agentic framework with specialized LLMs and automated debugging.

Manages a personal knowledge base by organizing notes as Markdown files directly on disk, providing a block-based editor, native relationships, and deep Git integration for version control, collaboration, and AI workflow support.

Equips AI agents with a comprehensive library of marketing skills, enabling them to perform conversion optimization, copywriting, SEO, and analytics tasks.

Manages AI agent context by sandboxing tool outputs and persistently tracking session state across multiple LLM development environments to prevent context window bloat.

Optimize AI agent communication by condensing LLM output and input tokens without losing technical accuracy.

Filters and compresses command outputs to significantly reduce LLM token consumption for developers using AI coding assistants.

Run a persistent, self-improving AI agent locally, accessible via Telegram, with support for any LLM provider.

Integrates 211 specialized AI expert roles into 16 programming tools, including 46 localized for China, and orchestrates multi-agent workflows for complex tasks.

Firecrawl Web Agent is an open-source framework and set of SDKs for building autonomous web research agents. It leverages Deep Agents (LangChain) for planning, acting, and observing, integrating web tools for scraping and interaction.

Optimizes AI Agent Skills using an autonomous, iterative loop, evaluating structural quality and real-world performance with a 'ratchet' mechanism to ensure continuous improvement.

MOSS-TTS-Nano is a 0.1B parameter, open-source multilingual text-to-speech model for realtime, CPU-friendly speech generation and voice cloning, supporting 20 languages with a simple local setup.

Enhances AI coding agents with a suite of skills to generate complex diagrams and data visualizations directly within Markdown, supporting software modeling, enterprise architecture, data analytics, and various content cards.

Hindsight is an open-source agent memory system designed to enable AI agents to learn and retain information over time, offering state-of-the-art performance in long-term memory tasks beyond simple conversation recall.

AI tool that generates professional architecture diagrams from plain text descriptions using Claude.ai, outputting a self-contained HTML file for easy sharing and iteration.