Bayesian control for coding agents

• URL: http://arxiv.org/abs/2606.24453v1
• Date: Tue, 23 Jun 2026 11:41:32 GMT
• Title: Bayesian control for coding agents
• Abstract summary: We propose a cost-sensitive sequential hypothesis testing framework for coding agents.<n>A Bayesian controller maintains a belief over correctness and decides whether to gather more evidence, refine the candidate, verify it, or stop.<n>We show that the belief state yields an interpretable correctness score that outperforms token-probability and raw tool-success baselines for uncertainty quantification.
• Score: 63.64172141184361
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Modern coding agents pair LLM generators with various tools, including cheap diagnostics and expensive verifiers. The tool-use decisions are typically governed by orchestrators that often use fixed rules and ignore uncertainty. We formulate orchestration as cost-sensitive sequential hypothesis testing: a Bayesian controller maintains a belief over candidate correctness and dynamically decides whether to gather more evidence, refine the candidate, verify it, or stop. Across six generators and nine coding benchmarks, Bayesian control proves to be most valuable when verification is costly and critics are informative but imperfect. Beyond control, the belief state yields an interpretable correctness score that outperforms token-probability and raw tool-success baselines for uncertainty quantification.

Related papers

• What Do Compliance Detectors Read? An Audit of Activation Probes and Guard Models [3.3598755777055374]
  Internal Compliance Score (ICS) is a training-free activation readout from ten labelled pairs and scored by a single projection.<n>We show this condition fails across the current class of compliance detectors, a failure we call rule blindness.<n>We release the counterfactual protocol and crossed-rule benchmark so rule blindness can be tested in future probe and guard claims.
  arXiv  Detail & Related papers  (2026-08-17T17:37:07Z)
• TrustChain-Review: A Risk-Adaptive Blockchain and Game-Theoretic Framework for Trustworthy AI-Assisted Code Review [0.0]
  This study introduces TrustChain-Review, a framework that combines verifiable evidence, strategic incentives, and risk-sensitive governance.<n>Six governance configurations are compared over 30 independent runs using reputation accuracy, trust convergence, malicious-review detection, superficial-review detection, net platform utility, governance cost, and cost-efficiency.<n>The full-evidence configuration produces the strongest reputation, trust, and detection results, but it also has the highest governance cost.
  arXiv  Detail & Related papers  (2026-07-29T17:53:52Z)
• Bet on Features: Anytime-Valid and Feature-Aware Auditing of Conditional Quantile Forecasters [4.458834881246723]
  Black-box conditional quantile forecasts are widely used for sequential decisions under asymmetric costs.<n>Forecasters can look to an auditor with coarse information while being miscalibrated to an auditor with richer information.<n>We develop a distribution-free and game-theoretic testing framework for continuously auditing black-box conditional quantile forecasters with non-i.i.d. losses.
  arXiv  Detail & Related papers  (2026-07-13T15:04:24Z)
• Exploring Agentic Tool-Calling Decisions via Uncertainty-Aligned Reinforcement Learning [63.24624171679711]
  Large language model (LLM)-based agents often make suboptimal tool-use decisions.<n>We propose TRUST, which incorporates uncertainty quantification into reward design as a repulsive force for maintaining uncertainty separation.
  arXiv  Detail & Related papers  (2026-06-05T07:08:34Z)
• The Hidden Signal of Verifier Strictness: Controlling and Improving Step-Wise Verification via Selective Latent Steering [67.8271652641864]
  We study whether verifier strictness can be controlled through hidden-state intervention.<n>VerifySteer exploits latent correctness signals for sample-level routing and selectively intervenes on paragraph boundaries.
  arXiv  Detail & Related papers  (2026-05-20T05:48:16Z)
• Bayesian Sequential Verification for Budget-Aware Quantum Program Testing [5.589961715298686]
  This paper formulates Bayesian sequential verification as a reference-based hypothesis testing workflow.<n>It is evaluated in Qiskit on two complementary workloads: Bell-state and QAOA-MaxCut.
  arXiv  Detail & Related papers  (2026-05-15T04:21:08Z)
• IMMACULATE: A Practical LLM Auditing Framework via Verifiable Computation [49.796717294455796]
  We present IMMACULATE, a practical auditing framework that detects economically motivated deviations.<n>IMMACULATE selectively audits a small fraction of requests using verifiable computation, achieving strong detection guarantees while amortizing cryptographic overhead.
  arXiv  Detail & Related papers  (2026-02-26T07:21:02Z)
• When to Trust the Cheap Check: Weak and Strong Verification for Reasoning [26.38833436936642]
  We formalize the tension between strong and weak verification.<n>We show that optimal policies admit a two-threshold structure and that calibration and sharpness govern the value of weak verifiers.<n>We develop an online algorithm that provably controls acceptance and rejection errors without assumptions on the query stream, the language model, or the weak verifier.
  arXiv  Detail & Related papers  (2026-02-19T18:47:38Z)
• Preventing the Collapse of Peer Review Requires Verification-First AI [49.995126139461085]
  We propose truth-coupling, i.e. how tightly venue scores track latent scientific truth.<n>We formalize two forces that drive a phase transition toward proxy-sovereign evaluation.
  arXiv  Detail & Related papers  (2026-01-23T17:17:32Z)
• E-valuator: Reliable Agent Verifiers with Sequential Hypothesis Testing [7.984665398116918]
  We introduce e-valuator, a method to convert any black-box verifier score into a decision rule with provable control of false alarm rates.<n>Evaluator builds on tools from eprocesses to develop a sequential hypothesis test that remains statistically valid at every step of an agent's trajectory.<n>We show that e-valuator provides greater statistical power and better false alarm rate control than other strategies across six datasets and three agents.
  arXiv  Detail & Related papers  (2025-12-02T05:59:18Z)
• Validating Solidity Code Defects using Symbolic and Concrete Execution powered by Large Language Models [0.0]
  This paper introduces a novel detection pipeline that integrates custom Slither-based detectors, Large Language Models (LLMs), Kontrol, and Forge.<n>Our approach is designed to reliably detect defects and generate proofs.
  arXiv  Detail & Related papers  (2025-09-16T12:46:11Z)
• Trusted Uncertainty in Large Language Models: A Unified Framework for Confidence Calibration and Risk-Controlled Refusal [31.458406135473805]
  We present UniCR, a unified framework that turns heterogeneous uncertainty evidence into a calibrated probability of correctness.<n>UniCR learns a lightweight calibration head with temperature scaling and proper scoring.<n>Experiments on short-form QA, code generation with execution tests, and retrieval-augmented long-form QA show consistent improvements in calibration metrics.
  arXiv  Detail & Related papers  (2025-09-01T13:14:58Z)
• COIN: Uncertainty-Guarding Selective Question Answering for Foundation Models with Provable Risk Guarantees [51.5976496056012]
  COIN is an uncertainty-guarding selection framework that calibrates statistically valid thresholds to filter a single generated answer per question.<n>COIN estimates the empirical error rate on a calibration set and applies confidence interval methods to establish a high-probability upper bound on the true error rate.<n>We demonstrate COIN's robustness in risk control, strong test-time power in retaining admissible answers, and predictive efficiency under limited calibration data.
  arXiv  Detail & Related papers  (2025-06-25T07:04:49Z)

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.