PREPING: Building Agent Memory without Tasks

2026-05-11

AI Summary

This work addresses the cold-start problem faced by agents in novel environments due to a lack of prior task experience by proposing a pre-task memory construction method that operates without real tasks. The approach introduces a Proposer-guided framework comprising three collaborative modules—Proposer, Solver, and Validator—to automatically generate structured synthetic tasks, filter feasible trajectories, and selectively store them into procedural memory, thereby avoiding redundant and unproductive exploration. Experimental results demonstrate that the method significantly outperforms memory-free baselines on AppWorld, BFCL v3, and MCP-Universe, achieving performance comparable to strong baselines that rely on offline or online experience while reducing deployment costs by factors of 2.99 and 2.23, respectively.

Abstract

Agent memory is typically constructed either offline from curated demonstrations or online from post-deployment interactions. However, regardless of how it is built, an agent faces a cold-start gap when first introduced to a new environment without any task-specific experience available. In this paper, we study pre-task memory construction: whether an agent can build procedural memory before observing any target-environment tasks, using only self-generated synthetic practice. Yet, synthetic interaction alone is insufficient, as without controlling what to practice and what to store, synthetic tasks become redundant, infeasible, and ultimately uninformative, and memory further degrades quickly due to unfiltered trajectories. To overcome this, we present Preping, a proposer-guided memory construction framework. At its core is proposer memory, a structured control state that shapes future practice. A Proposer generates synthetic tasks conditioned on this state, a Solver executes them, and a Validator determines which trajectories are eligible for memory insertion while also providing feedback to guide future proposals. Experiments on AppWorld, BFCL v3, and MCP-Universe show that Preping substantially improves over a no-memory baseline and achieves performance competitive with strong playbook-based methods built from offline or online experience, with deployment cost $2.99\times$ lower on AppWorld and $2.23\times$ lower on BFCL v3 than online memory construction. Further analyses reveal that the main benefit does not come from synthetic volume alone, but from proposer-side control over feasibility, redundancy, and coverage, combined with selective memory updates.

Problem

• agent memory
• cold-start problem
• pre-task memory construction
• synthetic practice
• procedural memory

Innovation

• pre-task memory construction
• proposer-guided framework
• synthetic practice
• selective memory update
• agent memory