Evolver, the core engine of the EvoMap ecosystem, is a GEP-protocol-powered self-evolution engine for AI agents. It transforms ad-hoc prompt adjustments into auditable and reusable evolution assets like Genes and Capsules. By scanning runtime logs and identifying error patterns, Evolver selects suitable assets and generates standardized GEP prompts to guide the AI agent's next evolutionary step. It supports various strategy presets and integrates with major agent runtimes like OpenClaw and Cursor, enabling automated agent self-repair and capability enhancement.
OpenHarness, developed by HKUDS, is an open-source, lightweight Python framework providing essential infrastructure for advanced AI Agents. Functioning as a comprehensive 'Agent Harness,' it equips large language models with robust tool-use capabilities (43+ tools), skill management, persistent memory, and sophisticated multi-agent coordination. Key features include parallel streaming agent loops, intelligent context compression, multi-level permission governance, and a flexible plugin ecosystem. OpenHarness empowers researchers and developers to deeply understand, experiment with, and extend production-grade AI Agents. Its capabilities are showcased by ohmo, an intelligent assistant for multi-day tasks like code generation and PR management across platforms such as Feishu and Slack, leveraging existing LLM subscriptions without extra API keys.
Open Multi-Agent is a lightweight, TypeScript-native multi-agent orchestration framework. By simply providing a goal, its built-in coordinator automatically decomposes the goal into a task DAG, parallelizes independent steps, and synthesizes the final output. With only 3 runtime dependencies, it seamlessly drops into any Node.js backend. Key features include mixing over 12 LLM providers within a single team, a sandboxed filesystem workspace, pluggable shared memory, and native integration with the Model Context Protocol (MCP) for tool extensibility.
DeepSeek-Reasonix by esengine is an open-source, DeepSeek-native AI coding agent built for the terminal and desktop. Its core innovation is a cache-first architecture engineered around prefix-cache stability. By maintaining an append-only loop, it achieves near 99.8% cache hit rates during extensive coding sessions, slashing API token costs by up to 5x. Alongside its robust CLI, Reasonix features a native Tauri desktop GUI and remote QQ channel integration. It fully supports the Model Context Protocol (MCP) for tool expansion, local semantic indexing, web search, and custom Markdown-based skills, making it a highly optimized companion for developers.