Jun 2026· arXiv.org· Vol abs/2606.31763· 0 citations· 16 references
Computer Science
TL;DR
Results show that the evaluation framework captures execution-relevant requirements for autonomous wet-lab automation, and that ProtoPilot can meet them by converting protocol and code generation into validated execution and feedback-guided revision.
Abstract
Autonomous wet-lab experimentation requires more than plausible protocol text: biological intent, quantitative procedures, device constraints and experimental feedback must remain aligned from protocol and SOP design to code and physical execution. We developed ProtoPilot, a self-evolving multi-agent system, together with an expert-grounded benchmark and evaluation framework for testing this conversion as an experimental automation problem. The framework spans 294 synthetic-biology and molecular-biology tasks derived from 98 gold-standard protocols, wet-lab expert rubrics, device-level validity gates and real experimental tests. ProtoPilot incorporates layer-wise verifiability, multi-agent orchestration and a runtime-updated skill library to generate protocols, expand SOPs, synthesize SDK-compliant code and revise workflows from wet-lab feedback. It achieved a Top@3 expert-preference rate of 90.2%, an overall protocol-to-code gate pass rate of 89.5% and an Opentrons pass rate of 88.24%, compared with 32.35% for OpenTrons-AI. Wet-lab validation produced interpretable readouts, Sanger-confirmed products and feedback-corrected PCA-assembled DNA targets, establishing a verifiable route to autonomous experimentation. Together, these results show that the evaluation framework captures execution-relevant requirements for autonomous wet-lab automation, and that ProtoPilot can meet them by converting protocol and code generation into validated execution and feedback-guided revision.
VISA is presented, a structured, symbol-based description protocol that specifies a model in eight interconnected tables---four at the agent level (Agent, Variable, Sensing, Internal Function) and four at the model level (Associated Data, Input/Output, Schedule, Validation)---under the principle of minimality with completeness.
TestAgent is proposed, an LLM-based test generation approach that addresses the above limitations by emulating human testing practices via a multi-agent collaboration mechanism and equips TestAgent with a set of tool APIs that can be invoked dynamically in an on-demand and adaptive manner.
Quanjun Zhang, Ye Shang, Siqi Gu et al.· 0 citations
The evolution from reactive to proactive AI systems represents a paradigm shift in software engineering, enabling autonomous agents to anticipate requirements, plan complex workflows, and execute multi-step development tasks without human intervention. This paper presents a novel multi-agent architecture for autonomous code generation and software maintenance in enterprise environments. Our framework integrates specialized AI agents for requirements analysis, code synthesis, testing, and deployment, coordinated through a hierarchical planning system with adaptive decision-making capabilities. The system employs reinforcement learning to optimize task allocation, learn from code review feedback, and improve over time. Experimental evaluation on real-world enterprise codebases demonstrates that our proactive agent system achieves 87.3% automated issue resolution, reduces bug fix latency by 62%, and maintains code quality metrics comparable to human developers. The framework successfully handles complex refactoring tasks, security vulnerability remediation, and feature implementation with minimal human oversight, representing a significant advancement toward fully autonomous software engineering workflows.
Saher Elsayed, Samer Abubaker, M. Ali et al.· Annual International Compute...· 0 citations
EvoSOP is introduced, a framework that empowers agents to extract SOPs from execution trajectories and iteratively optimize the toolset through a systematic lifecycle of construction, merging, evaluation, and pruning, providing a scalable pathway for the development of self-evolving agents.
Haipeng Ding, Yuexiang Xie, Zhewei Wei et al.· 2 citations
Large Language Model (LLM) based agents exhibit substantial run-to-run execution variance even when given identical tasks and tools -- acceptable for exploratory use but unacceptable in regulated domains such as finance and compliance. We study harness engineering: wrapping an agent in a deterministic execution layer (finite-state control, forced tool selection, output validation, bounded retry, and structured planning) and measuring its effect on execution determinism and task success. Across two synthetic tasks (finance and legal) and two open-weight models (Qwen-2.5-7B-Instruct, Gemma-3-27B), a first-pass harness produces a mixed result: it significantly improves reproducibility in one of four model-task cells, significantly degrades it in two, and has no effect in the fourth. A trace-level diagnostic finds the cause: once tool sequence, state sequence, and output are already highly consistent, an unconstrained free-text planning step becomes the dominant remaining source of variance. Adding Structured Planning -- validating the plan against a fixed schema before any tool is invoked -- eliminates the effect entirely: three of four cells reach a Reproducibility Rate and Determinism Index of 1.000 at N=100, and task success rises to 100% in three of four cells. The gain is not free: token cost falls in every cell, but latency shows a genuine, sample-size-robust split by model -- one becomes faster under the constraint, the other markedly slower. We argue harness engineering is a distinct, effective discipline for agent reliability, but its cost is model-dependent and must be measured, not assumed.