Self Improvement Python Packages

areev

Self-improving agents, governed. Areev is the substrate for adaptive agents — agents that get better from their own history, under human authority, in steps you can inspect, undo, and re-measure.

agentdescent

Gradient descent, but the parameters are agents — a parallel, asynchronous framework for self-evolving agents (skills, prompts, harnesses). Diffs are the gradients; the aggregator is the optimizer.

poppy-ai

Poppy — continuous skill mining for coding agents: scheduled agents turn session history into human-reviewed, evidence-backed skills

mnfst

Manifest SDK for Python. See every API call your app makes, heal the ones that fail, and know what will break next.

seed-evolution

Self-evolving coding-agent harness: immutable Rust guardian judges an evolvable TS/Python organism (Pareto promotion, friction-driven lab).

personagent

A group-chat character that knows when to stay quiet and learns from being corrected, with an evidence ledger, rollback and a measured eval. QQ, Telegram, Discord and more through AstrBot, Koishi or Matrix.

darwin-memo

Self-curating memory for LLM agents: MeMo-style external memory kept honest by survival-based selection. No reward models, no judges.

simit

SIMIT: Self-Improving Vision-Language Models via Imagination at Test-Time, official pip library of the paper, SIMIT-ICL: vision-language models that imagine their own in-context demonstrations at test time

errlore

Memory for AI agents that learns from failures — lessons, known-issues injection, per-model trust. Embedded, file-based, no server.

adaptible

A small language model that keeps its conversations and retrains itself on its own corrections, on a laptop.

rsikit

Easy, clean, and transparent toolkit for developing and evaluating self-improving optimizers for a variety of tasks.

areev-crewai

Self-improving agents, governed. Areev is the substrate for adaptive agents — agents that get better from their own history, under human authority, in steps you can inspect, undo, and re-measure.

repocourse

AI 项目复盘助手：扫描代码项目，基于多源证据 + LLM 分析，生成《README复盘.md》（8 大板块）

kairos-agent

kairos picks the right LLM technique for any task by querying a curated wiki of agent patterns - RAG, ReAct, Reflexion, and 17 more. Karpathy LLM Wiki + rule-based selector + 3 working runners. Zero API keys (routes through llm-mcp).

memary

The Open Source Memory Layer For Autonomous Agents

areev-langgraph

Self-improving agents, governed. Areev is the substrate for adaptive agents — agents that get better from their own history, under human authority, in steps you can inspect, undo, and re-measure.

three-layer-agent-memory

A filesystem-native, version-controllable long-memory paradigm for AI agents. Surface/Middle/Deep layers + recall/writeback/consolidate protocol. Pure Markdown, zero infra, drives agent self-evolution.

wlpworld

A filesystem-native, version-controllable long-memory paradigm for AI agents. Surface/Middle/Deep layers + recall/writeback/consolidate protocol. Pure Markdown, zero infra, drives agent self-evolution.

spellwell

Practice spelling using your terminal.

ratchet-memory

AI agents that only get better.

ratchet-research

AI agents that only get better.

ratchet-core

AI agents that only get better.

ratchet-pilot

AI agents that only get better.

ratchet-factory

AI agents that only get better.