Policy-as-Skill: Governed LLM Decision Support with Evidence, Deterministic Control, and Audit

Kabeh Mohsenzadegan; Vahid Tavakkoli; Kyandoghere Kyamakya

In the authors' words

Organizations increasingly use LLMs for policy, compliance, risk, and operational decision support, requiring evidence validation, review routing, version control, and auditability. We introduce Policy-as-Skill (PaS), a modular runtime that packages these functions as executable, versioned policy capabilities. Thirteen methods are evaluated with a fixed Gemma4 backend on 600 development tasks. PaS+Audit achieves 53.8% exact accuracy, macro-F1 0.346, review F1 0.854, citation precision 1.000, policy-reference recall 0.984, and audit completeness 1.000, outperforming LLM+RAG on most governance and review metrics. Deterministic control raises aggregate accuracy to 61.2% but is strongly task dependent, supporting selective rather than universal rule-based intervention.

Main resultLimitation the authors admit

Policy-as-Skill: Governed LLM Decision Support with Evidence, Deterministic Control, and Audit

In the authors' words

More on Language models that day

Beyond Poetry: Can Large Language Models Generate Classical Arabic Maqamat?

EviStreams: Human-in-the-Loop AI Data Extraction for Systematic Reviews in Medicine

Realize What Matters: Principled Context Representation for Large-Scale Reasoning

COMED: The Missing Middle Between Routing and Collaboration in Multi-LLM Inference

Repurposing Pre-trained LLMs as High Fidelity Continuous Text Autoencoders