PrologMCP: A Standardized Prolog Tool Interface for LLM Agents
Agnieszka Mensfelt, Adarsh Prabhakaran, Adrian Haret, Vince Trencsenyi, Kostas Stathis
Why It Matters
What makes this one worth your time
This work addresses the limitations of current LLMs in deductive reasoning, providing a practical solution that can be widely adopted in AI applications requiring logical inference.
PrologMCP offers a robust interface for enhancing LLMs with Prolog's reasoning capabilities.
Summary
The paper introduces PrologMCP, an open-source server that provides a standardized interface for integrating Prolog with language models, enabling improved performance on deductive reasoning tasks.
Key contributions
- Introduction of PrologMCP as a task-agnostic tool interface for LLMs.
- Demonstration of performance improvements in deductive reasoning tasks compared to standard LLMs.
- Provision of structured error reporting and session isolation for enhanced usability.
Notable insights
- The use of the Model Context Protocol (MCP) allows for a structured and reusable approach to integrating Prolog with LLMs.
- The evaluation against standard reasoning LLMs highlights the potential of symbolic delegation in improving performance on challenging reasoning tasks.
Possible limitations
- Not stated in the abstract.
Abstract
arXiv:2606.14935v1 Announce Type: new Abstract: Frontier reasoning-tuned language models still fail on deductive tasks at depth, and the cost of improved performance through extended internal reasoning scales poorly. Symbolic delegation offers a complementary route: a language model translates the problem, while a solver performs the inference. However, current autoformalization pipelines for logic programming are typically bespoke integrations tied to particular tasks or agents. We introduce PrologMCP, a task-agnostic, open-source server that exposes Prolog as a stateful tool through the Model Context Protocol (MCP). Its compact tool interface, structured error reporting, and per-session isolation make the translate-run-inspect-repair loop a reusable primitive for MCP-capable agents. We evaluate a formalizer agent enhanced with PrologMCP against standard and reasoning LLMs (Claude Sonnet 4.6, GPT-4.1, and o4-mini) on two subsets of PARARULE-Plus: a general-purpose sample and a more challenging one targeting a specific failure mode of natural-language reasoning. On the general sample, the formalizer matches or exceeds reasoning LLMs (accuracy 1.00 vs.\ 1.00 / 0.998), with the largest gains over standard models (0.762 for GPT-4.1). On the challenging subset, the formalizer remains near-perfect (1.00 / 0.99) while reasoning LLMs drop to 0.95 / 0.94. These results suggest that delegating inference to Prolog via MCP is a robust and inspectable alternative to extended natural-language reasoning.