OpenAkita - an open-source AI assistant framework with self-evolution capabilities.
OpenAkita is an open-source, self-evolving AI assistant framework. It can execute commands like a traditional AI assistant, possessing a unique 'self-evolution' capability that automatically performs memory consolidation, error correction, and task review. When it encounters a bottleneck, it can automatically...
What is OpenAkita?
OpenAkita is an open-source, self-evolving AI assistant framework. Like traditional AI assistants, it executes commands and possesses a unique "self-evolution" capability, automatically performing memory consolidation, error repair, and task review. When encountering obstacles, it can automatically generate new skills and install dependencies, truly becoming smarter with use. The framework includes 8 built-in personality profiles (girlfriend, butler, Jarvis, etc.), supporting proactive greetings and remembering user preferences, making interactions more "present." Technically, it supports dynamic hot-switching and automatic failover across 9+ LLMs, covering 7 major IM platforms including Telegram, Lark, DingTalk, and QQ, and may be the first AI agent to support "meme battles."
Main functions of OpenAkita
-
Self-learning and evolutionIt automatically performs daily memory consolidation, error self-check and repair, and task review. When it encounters a bottleneck, it automatically generates new skills and installs dependencies, becoming smarter the more you use it.
-
8 types of character designsIt offers 8 preset personalities, including girlfriend, butler, and Jarvis, and supports features such as proactive greetings, remembering user preferences, and automatic muting at night to enhance user presence.
-
Plan ModeComplex tasks are automatically broken down into multi-step plans, and the execution progress is tracked in real time, forming a closed loop of "planning → execution → verification".
-
Dynamic multiple LLM supportSupports hot switching between 9+ model providers, with priority routing and automatic failover; seamlessly switches to the next provider if one fails.
-
Skills Market and MCP StandardsSupports Agent Skills / MCP open standards, allowing you to install skills from GitHub with one click and use them immediately.
-
Integration with 7 major IM platformsSupports Telegram, Lark, WeChat Work, DingTalk, QQ official robot, OneBot, and CLI command line.
-
Emoji featureIt may be the first AI Agent to support "meme battles," with over 5,700 built-in emoji packs that automatically match based on mood and personality.
OpenAkita's technical principles
-
Layered architecture designIt adopts a four-layer architecture: desktop (Tauri + React) + core layer (Brain/Memory/Ralph) + tool layer + evolution layer, to achieve decoupling of interface, logic, tools and evolution.
-
Ralph Never Give Up the CycleThe core execution engine is responsible for the persistent execution and error recovery of tasks, ensuring that tasks are automatically retried or new skills are generated to resolve them when they get stuck.
-
Vector memory systemIt stores long-term memory based on a vector database, supports contextual retrieval and memory integration, and enables cross-session user preference memory.
-
Prompt compilerDynamically compile multi-turn dialogue contexts and tool call instructions to optimize LLM input/output efficiency.
-
Active engine mechanism: Trigger proactive behaviors based on time and events (such as scheduled greetings or nighttime silencing) to give the Agent a sense of "presence".
-
MCP protocol supportsIt follows the Model Context Protocol open standard to achieve standardized integration with external tools.
-
Skills are automatically generatedWhen encountering a task that cannot be completed, it automatically generates Python skill code and installs dependencies, thus achieving self-extension.
-
Multi-LLM routing policyIt dynamically selects model providers based on priority and availability, and supports automatic failover and load balancing.
-
Daily Evolution PipelineIt performs tasks such as memory consolidation, error log analysis, and skill optimization on a regular basis, forming a "daily update" self-iteration.
OpenAkita project address
- Project official websitehttps://openakita.ai/
- Github repositoryhttps://github.com/openakita/openakita
Application scenarios of OpenAkita
-
Personal intelligent assistantAs a daily AI assistant, it handles tasks such as schedule management, information retrieval, and file organization, supports multi-platform message synchronization, and becomes more adept at understanding user habits the more it is used.
-
Multi-platform customer service robotDeployable to platforms such as Telegram, Lark, WeChat Work, DingTalk, and QQ, providing 24/7 automatic replies and intelligent customer service.
-
Automated office assistantIt combines planning mode to break down complex workflows and automatically execute repetitive office tasks such as email processing, report generation, and data organization.
-
Emotional companionship interactionThrough eight personality settings, including girlfriend and housekeeper, it provides an emotional companionship experience with features such as proactive greetings, memory preferences, and emoji battles.
-
Development and testing toolsDevelopers can leverage its MCP standard and skills marketplace to quickly integrate custom tools and use it as a framework for AI Agent development for further development.
-
Smart Home HubIt leverages tools such as Shell, browsers, and desktop automation to control local devices and smart home systems.