Procedural Graphs Enable Self‑Evolving Execution Structures for LLM Agents
Large language models are increasingly used as autonomous agents that plan over long horizons and call external tools. Existing agents typically generate actions freely from an ever‑growing history, which leaves the procedural steps—what to do, in what order, and under which conditions—implicit. This can cause agents to lose sight of goals, invoke tools out of sequence, or repeat ineffective…
Key points
- Procedural Graphs represent procedural knowledge as (procedure, relation, procedure) triples for LLM agents
- The graph self‑evolves by comparing failed and successful trajectories, editing its topology to improve performance
- Experiments across datasets and LLMs show consistent gains over memory‑based baselines without manual engineering
The new framework introduces a Procedural Graph that mirrors a knowledge graph but for procedures, encoding (procedure, relation, procedure) triples. At each decision point the system identifies the active node and a guidance model translates the surrounding subgraph into situational hints that bias the next action without dictating it. Crucially, the graph evolves itself: an LLM refiner compares failed and successful runs, edits the graph’s topology and attributes, and retains edits that improve validation performance while preserving rejected paths to discourage repetition.
Across several benchmark datasets, task types, and LLM back‑ends, the Procedural Graph consistently outperforms memory‑based baselines, and its self‑evolution further lifts performance without any hand‑crafted engineering. The approach also demonstrates the ability to repair flawed expert priors, suggesting a path toward more reliable, self‑improving AI agents.
Procedural Graphs: Self-Evolving Execution Structures for LLM Agents
arxiv.org · 9 September 2026
Computer Science > Artificial Intelligence
Title:Procedural Graphs: Self-Evolving Execution Structures for LLM Agents
View PDF HTML (experimental) Abstract:Large language models are increasingly deployed as agents that plan over long horizons and act through external tools. Most agents select actions through unconstrained generation over an accumulating history, leaving implicit the procedural knowledge of what to do, in what order, and under which conditions. As trajectories lengthen, agents can lose track of their objectives, invoke tools out of order, and repeat unproductive actions. We introduce the Procedural Graph: just as a knowledge graph organizes factual knowledge into (entity, relation, entity) triplets for what-is questions, a Procedural Graph organizes procedural knowledge into (procedure, relation, procedure) triplets for what-to-do questions. At each decision step, the framework localizes the agent's active node, and a guidance model translates the surrounding subgraph into step-level situational guidance that biases the solver's next action without dictating it. The graph is self-evolving: an LLM refiner contrasts failed trajectories with successful ones and edits the graph's topology and attributes, committing edits that preserve or improve held-out validation performance while retaining rejected ones to discourage repetition. Starting from a minimal skeleton, the loop builds graphs that match or surpass hand-designed ones. It can also repair a flawed expert prior. Across multiple datasets, task types, and LLMs, the Procedural Graph delivers consistent gains over memory-based baselines, and self-evolution further improves performance without manual engineering.
Current browse context:
Bibliographic and Citation Tools
Code, Data and Media Associated with this Article
Demos
Recommenders and Search Tools
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
This text was published by arxiv.org . It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
Coverage and discussion
1 source- Hacker News discussion · 57 points news.ycombinator.com
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us.
Comments
via GitHub Discussions