thedotmack/claude-mem
Claude-Mem is a plugin for Claude Code that captures, compresses, and preserves coding session context to enable persistent memory and continuity across sessions.
Awesome AI Agents › Continual Learning Agents
YunjueTech/Yunjue-Agent
Yunjue Agent is the official implementation of a self-evolving agent system from Yunjue Technology, designed for open-ended tasks that a fixed toolset cannot cover. Its premise is that an agent should not depend on tools handed to it in advance: starting from a zero-start setting with no task-specific tooling, the system decomposes a task, writes the tools it needs in situ, executes them and keeps the useful ones, so its capability grows during the run rather than only during training. The implementation is Python, requires version 3.12 and uv, and installs the Codex CLI as the execution backend, which the operator configures separately with their own API key. Configuration is split between an environment file for keys and execution limits such as worker recursion depth and maximum task execution count, and a YAML file that selects the vision and summarisation models. Reproducibility is a stated goal of the release. A dedicated reproduce branch carries the evaluation scripts and workflow, full system logs under zero-start settings are published for five benchmarks including Humanity's Last Exam, DeepSearchQA, FinSearchComp, xbench-ScienceQA and xbench-DeepSearch, and the complete traces are also distributed as a Hugging Face dataset for analysis. A technical report on arXiv covers the theory, system cost and a metric the authors call Evolutionary Generality Loss, which measures how specialising toward one task degrades general capability. Two demos ship with the code: a web demo showing tool decomposition, where the agent writes tools to search and scrape PDFs or look up stock information, and a command line skill demo. The audience is researchers working on self-improving agents, tool synthesis and deep search evaluation.
https://github.com/YunjueTech/Yunjue-Agent
Also in Continual Learning Agents
vectorize-io/hindsight
Hindsight is an agent memory system designed to create smarter, learning AI agents, addressing limitations of traditional memory techniques like RAG and knowledge graphs.
langchain-ai/langchainjs
LangChain.js is a TypeScript framework for building context-aware, reasoning applications powered by large language models, offering modular components, chains, agents, and tools for production and deployment.
aden-hive/hive
Hive is an outcome-driven agent development framework designed for evolving agents with AI evaluation, human-in-the-loop, and self-improvement capabilities.
aiming-lab/MetaClaw
Proxy-based layer that turns everyday conversations with a personal agent into learning signals, injecting skills per turn and running scheduled reinforcement learning updates without a local GPU.
kayba-ai/agentic-context-engine
Agentic Context Engine (ACE) is a framework that enables AI agents to autonomously learn from their successes and failures, continuously improving their performance across various tasks without additional training data or fine-tuning.
AgentDock/AgentDock
AgentDock is an open-source framework for building sophisticated AI agents with configurable determinism, enabling reliable and creative AI applications through a node-based architecture and multi-stage workflows.
modelscope/AgentEvolver
Self-evolving agent training framework from ModelScope that combines automatic task generation, experience-guided exploration and attribution-based credit assignment into one reinforcement learning stack.