A Control System, a Dataset, and a Recipe for Making Frozen LLM Agents Learn a Domain

• URL: http://arxiv.org/abs/2607.25415v1
• Date: Tue, 28 Jul 2026 08:10:25 GMT
• Title: A Control System, a Dataset, and a Recipe for Making Frozen LLM Agents Learn a Domain
• Abstract summary: Two 2026 systems, Meta-Harness and HyperAgents, show that this harness can itself be optimized or even self-rewritten by an agentic proposer.<n>We take a narrower, more constrained position: treat the harness as a small, fixed, human-legible action space and learn a policy over it online.<n>We instantiate this control system with DSPy as both the context assembler and the source of the strongest non-adaptive baseline.
• Score: 2.452218069720666
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Production LLM agents are increasingly assembled from a frozen model wrapped in a harness: a prompt template, a tool set, a memory/retrieval layer, a planning strategy, and a verification policy. Two 2026 systems, Meta-Harness (Lee et al., 2026) and HyperAgents (Meta AI, 2026), show that this harness can itself be optimized or even self-rewritten by an agentic proposer -- at the cost of either an expensive code-search loop or unconstrained self-modifying code, neither of which is auditable or usable with a fully black-box model API. We take a narrower, more constrained position: treat the harness as a small, fixed, human-legible action space and learn a policy over it online with classic sample-efficient reinforcement learning (an $ε$-greedy contextual bandit and REINFORCE), scored against a multi-objective reward (task success, verifier score, policy compliance, cost, latency, and an unsupported-claim penalty). We instantiate this control system with DSPy (Khattab et al., 2024) as both the context assembler and the source of the strongest non-adaptive baseline (a DSPy BootstrapFewShot static prompt), and evaluate it across three verifiable task domains -- tool-use workflows, code generation (HumanEval), and multi-hop retrieval QA (HotpotQA) -- and two model providers (a local Ollama model and AWS Bedrock). We release the harness-control-system code, the cross-domain verifiable task suite, the full trajectory/reward-decomposition logs from training, and a provider-agnostic deployment recipe for applying this to a new organization's domain and verification setup.

Related papers

• Iris: Climbing to the Search Frontier [21.066541702697112]
  We present Iris-mini and Iris-pro, two search agents trained at the 35B-A3B and 397B-A17B scales, together with the data pipeline and training recipe behind them.
  arXiv  Detail & Related papers  (2026-09-03T17:51:01Z)
• Context Assembly as the Controlled Variable: A Control-Theoretic View of Harness Policies for Frozen LLM Agents [2.452218069720666]
  We do not claim to introduce control theory to LLM agents -- that ship has sailed.<n>Prior work controls tool selection, inter-agent message routing, or the agent's raw action stream.<n>This paper develops the formal decomposition of the online controller.
  arXiv  Detail & Related papers  (2026-07-28T08:07:06Z)
• ClawArena-Team: Benchmarking Subagent Orchestration and Dynamic Workflows in Language-Model Agents [61.184954205350984]
  ClawArena-Team is a benchmark of 41 multi-turn, multimodal, multi-agent scenarios spanning 258 evaluation rounds and 72 staged updates.<n>All scoring is execution-based with no LLM judge: an overall score -- the Subagent-Management Score (SMS) -- multiplies task correctness by a least-privilege and modality-routing factor.
  arXiv  Detail & Related papers  (2026-06-30T06:06:08Z)
• ClayBuddy: A Framework, Evaluation, & Mitigation of Coding Agent Failures [1.0742675209112622]
  We study failure modes as stemming from three mechanisms: underspecification, capability errors, and agent harness errors.<n>We propose ClayBuddy, a harness that molds to user preferences and can be modified by the model in-session to mitigate these errors.
  arXiv  Detail & Related papers  (2026-06-13T18:29:53Z)
• Code as Agent Harness [107.31925305395957]
  In emerging agentic systems, code is no longer only a target output.<n>Code increasingly serves as an operational substrate for agent reasoning, acting, environment modeling, and execution-based verification.<n>This survey provides a unified roadmap toward executable, verifiable, and stateful AI agent systems.
  arXiv  Detail & Related papers  (2026-05-18T17:59:03Z)
• ComplexMCP: Evaluation of LLM Agents in Dynamic, Interdependent, and Large-Scale Tool Sandbox [61.862814740220806]
  $textbfComplexMCP$ is a benchmark designed to evaluate agents in rigorous conditions.<n>Built on the Model Context Protocol (MCP), $textbfComplexMCP$ provides over 300 meticulously tested tools.
  arXiv  Detail & Related papers  (2026-05-11T16:20:51Z)
• HARBOR: Automated Harness Optimization [2.7324157162184157]
  We argue that harness design is a first-class machine-learning problem.<n>We give a reference solver, HARBOR, built from a block-additive SAAS surrogate, multi-fidelity cost-aware acquisition, and TuRBO trust regions.
  arXiv  Detail & Related papers  (2026-04-22T13:45:12Z)
• LLM Readiness Harness: Evaluation, Observability, and CI Gates for LLM/RAG Applications [51.56484100374058]
  We present a readiness harness for LLM and RAG applications that turns evaluation into a deployment decision workflow.<n>The system combines automated benchmarks, OpenTelemetry observability, and CI quality gates under a minimal API contract.<n>We evaluate the harness on ticket-routing and BEIR tasks with full Azure matrix coverage.
  arXiv  Detail & Related papers  (2026-03-28T18:03:32Z)
• GUI-360$^\circ$: A Comprehensive Dataset and Benchmark for Computer-Using Agents [59.107657859025586]
  GUI-360$circ$ is a large-scale, comprehensive dataset and benchmark suite designed to advance computer-using agents (CUAs)<n>The released corpus contains over 1.2M executed action steps across thousands of trajectories in popular Windows office applications.<n>The dataset supports three canonical tasks, GUI grounding, screen parsing, and action prediction, and a hybrid GUI+API action space.
  arXiv  Detail & Related papers  (2025-11-06T12:19:02Z)
• An agentic system with reinforcement-learned subsystem improvements for parsing form-like documents [0.0]
  We propose an agentic AI system that leverages Large Language Model (LLM) agents and a reinforcement learning driver agent to automate consistent, self-improving extraction.<n>Our work highlights the limitations of monolithic LLM-based extraction and introduces a modular, multi-agent framework with task-specific prompts.<n>This self-corrective adaptive system handles diverse documents, file formats, layouts, and LLMs, aiming to automate accurate information extraction without the need for human intervention.
  arXiv  Detail & Related papers  (2025-05-16T09:46:10Z)
• Collab: Controlled Decoding using Mixture of Agents for LLM Alignment [90.6117569025754]
  Reinforcement learning from human feedback has emerged as an effective technique to align Large Language models.<n>Controlled Decoding provides a mechanism for aligning a model at inference time without retraining.<n>We propose a mixture of agent-based decoding strategies leveraging the existing off-the-shelf aligned LLM policies.
  arXiv  Detail & Related papers  (2025-03-27T17:34:25Z)
• Scanflow: A multi-graph framework for Machine Learning workflow management, supervision, and debugging [0.0]
  We propose a novel containerized directed graph framework to support end-to-end Machine Learning workflow management. The framework allows defining and deploying ML in containers, tracking their metadata, checking their behavior in production, and improving the models by using both learned and human-provided knowledge.
  arXiv  Detail & Related papers  (2021-11-04T17:01:12Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.