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
A-EVO-Lab/a-evolve
A-Evolve is an open-source infrastructure for agentic evolution, the practice of automatically improving an agent instead of hand-tuning its prompts, tools and scaffolding. The stated interface is minimal: import the agent_evolve package, construct an Evolver with a path to a base agent and the name of a benchmark, and run a number of cycles; the system returns an improved agent. The design goal is to be domain-agnostic and algorithm-agnostic, so any agent can be evolved on any task suite using any evolution algorithm, with the framework supplying the harness, the evaluation loop and the bookkeeping that normally has to be written by hand for each new benchmark. The repository ships reference evolution algorithms and reports results obtained by applying them to a single frontier base model across a wide set of benchmarks, including MCP-Atlas, SWE-bench Verified and Lite, Terminal-Bench, SkillsBench, ARC-AGI, OSWorld, tau-bench, CL-Bench and WebArena-Infinity, with per-benchmark gains over the unevolved baseline listed in the README together with leaderboard positions. A companion technical report describes A-Evolve-Training, an autonomous post-training run in which the same loop trained a 30B Nemotron model over four rounds with no human in the loop and placed competitively on a public reasoning challenge leaderboard. The project is the official code release for the position paper arguing that agentic evolution is the path to evolving large language models, and both the position paper and the training report are linked on arXiv. It requires Python 3.11 or newer and is MIT licensed. The audience is researchers and practitioners working on self-improving systems, recursive self-improvement and continual learning for LLM agents, as well as teams that want an automated way to tune an existing agent against a benchmark.
https://github.com/A-EVO-Lab/a-evolve
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.