Synthesizing Procedural Memory: Challenges and Architectures in Automated Workflow Generation

TL;DR

This paper explores how large language models can autonomously generate executable procedural code for workflows, addressing key structural challenges to improve automation and scalability in agent-based systems.

Contribution

It introduces a scientific methodology for LLMs to synthesize robust, production-grade code skills autonomously, overcoming four key structural bottlenecks.

Findings

Identified four structural bottlenecks in automated skill generation.

Demonstrated autonomous code synthesis using a hypothesis-driven approach.

Improved scalability and robustness of generated workflows.

Abstract

While CodeMem establishes executable code as the optimal representation for agentic procedural memory, the mechanism for autonomously synthesizing this memory from a blank slate remains underexplored. This paper operationalizes the transition of Large Language Models from passive tool-users to active workflow architects. Through a high-fidelity case study of a cross-service orchestration task involving Outlook and OneDrive, we identify and address four structural bottlenecks in automated skill generation: the Discovery Gap involving navigation of large tool registries, the Verification Gap regarding grounding tool response structures, the Decomposition Gap which replaces inefficient search with Linear State Anchoring, and the Scaling Gap focused on concurrency and persistence. We demonstrate that by enforcing a scientific methodology of hypothesize, probe, and code, agents can autonomously write robust, production-grade code skills.