---
title: "Zhejiang's Memp Lets AI Agents Inherit Skills, Cutting Steps by 37%"
subtitle: "A new framework gives LLM agents a growing library of how-to knowledge, cutting steps and boosting accuracy on long-horizon tasks"
url: "https://alphasignal.ai/news/zhejiang-s-memp-lets-ai-agents-inherit-skills-cutting-steps-by-37"
author: "AlphaSignal Newsroom"
author_url: "https://alphasignal.ai/editorial-team"
company: "AlphaSignal"
company_url: "https://alphasignal.ai/company/alphasignal"
published: "2026-04-15T00:00:00.000Z"
updated: "2026-08-23T21:25:37.978Z"
type: "Paper"
topics:
  - "Agents"
  - "Post Training"
subtopics:
  - "Memory"
  - "Fine Tuning"
  - "Distillation"
upvotes: 4500
source_url: "https://arxiv.org/abs/2508.06433"
comments: 0
truncated: true
---

# Zhejiang's Memp Lets AI Agents Inherit Skills, Cutting Steps by 37%

> A new framework gives LLM agents a growing library of how-to knowledge, cutting steps and boosting accuracy on long-horizon tasks

LLM agents keep re-solving the same problems from scratch. Every time one tackles a multi-step task, whether booking travel with tool calls or navigating a household simulator, it burns tokens rediscovering things it already figured out on a previous run. A team from Zhejiang University and Alibaba has proposed a fix called [Memp](https://arxiv.org/abs/2508.06433) that treats an agent's step-by-step know-how as something you can build, retrieve, update, and even transfer between models.

## The brittle memory problem

Contemporary agents built on LLMs can compose short action plans or call external tools, but their procedural knowledge is either hand-crafted, stored as fragile prompt templates, or implicitly entangled in model parameters that are expensive to update. Frameworks like LangGraph, AutoGPT, and cognitive architectures such as Soar offer memory buffers or rule chunks, but nobody has systematically studied how skills should be stored, indexed, patched, and retired over an agent's lifetime.

Procedural memory, in the human sense, is the automatic knowledge behind skills like typing or riding a bike. For an agent, the equivalent is the reusable pattern that says "to heat an egg, first find the fridge, then the microwave, then run these commands in this order." Without it, a fresh task means fresh trial and error, and on unfamiliar problems the agent burns many actions and tokens just understanding the environment before it can act.

_Preview only. Full text requires an AlphaSignal Pro account._

## Sources

- [Memp: Exploring Agent Procedural Memory](https://arxiv.org/abs/2508.06433)
- [MemP GitHub repository](https://github.com/zjunlp/MemP)
- [TravelPlanner benchmark (ICML 2024)](https://openreview.net/forum?id=l5XQzNkAOe)
- [ALFWorld (ICLR 2021)](https://openreview.net/forum?id=0IOX0YcCdTn)