Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

G2CP: A Graph-Grounded Communication Protocol for Verifiable and Efficient Multi-Agent Reasoning

Karim Ben Khaled, Davy Monticolo

TL;DR

G²CP addresses semantic drift, hallucinations, and token inefficiency in language-driven multi-agent systems by grounding inter-agent communication in graph operations over a shared knowledge graph. It formalizes a graph-based agent communication language with deterministic semantics, auditability, and commitment-based semantics, and demonstrates an architecture with four specialized agents guided by LLM-driven operation selection. Empirical results on 500 synthetic and 21 real industrial cases show a 73% reduction in inter-agent tokens and a 34% accuracy improvement over free-text baselines, with complete elimination of hallucinations and fully auditable reasoning traces. The work provides theoretical guarantees (determinism, completeness, security) and practical guidance for deploying graph-grounded MAS in safety-critical domains, with public release of code, data, and evaluation scripts.

Abstract

Multi-agent systems powered by Large Language Models face a critical challenge: agents communicate through natural language, leading to semantic drift, hallucination propagation, and inefficient token consumption. We propose G2CP (Graph-Grounded Communication Protocol), a structured agent communication language where messages are graph operations rather than free text. Agents exchange explicit traversal commands, subgraph fragments, and update operations over a shared knowledge graph, enabling verifiable reasoning traces and eliminating ambiguity. We validate G2CP within an industrial knowledge management system where specialized agents (Diagnostic, Procedural, Synthesis, and Ingestion) coordinate to answer complex queries. Experimental results on 500 industrial scenarios and 21 real-world maintenance cases show that G2CP reduces inter-agent communication tokens by 73%, improves task completion accuracy by 34% over free-text baselines, eliminates cascading hallucinations, and produces fully auditable reasoning chains. G2CP represents a fundamental shift from linguistic to structural communication in multi-agent systems, with implications for any domain requiring precise agent coordination. Code, data, and evaluation scripts are publicly available.

G2CP: A Graph-Grounded Communication Protocol for Verifiable and Efficient Multi-Agent Reasoning

TL;DR

G²CP addresses semantic drift, hallucinations, and token inefficiency in language-driven multi-agent systems by grounding inter-agent communication in graph operations over a shared knowledge graph. It formalizes a graph-based agent communication language with deterministic semantics, auditability, and commitment-based semantics, and demonstrates an architecture with four specialized agents guided by LLM-driven operation selection. Empirical results on 500 synthetic and 21 real industrial cases show a 73% reduction in inter-agent tokens and a 34% accuracy improvement over free-text baselines, with complete elimination of hallucinations and fully auditable reasoning traces. The work provides theoretical guarantees (determinism, completeness, security) and practical guidance for deploying graph-grounded MAS in safety-critical domains, with public release of code, data, and evaluation scripts.

Abstract

Multi-agent systems powered by Large Language Models face a critical challenge: agents communicate through natural language, leading to semantic drift, hallucination propagation, and inefficient token consumption. We propose G2CP (Graph-Grounded Communication Protocol), a structured agent communication language where messages are graph operations rather than free text. Agents exchange explicit traversal commands, subgraph fragments, and update operations over a shared knowledge graph, enabling verifiable reasoning traces and eliminating ambiguity. We validate G2CP within an industrial knowledge management system where specialized agents (Diagnostic, Procedural, Synthesis, and Ingestion) coordinate to answer complex queries. Experimental results on 500 industrial scenarios and 21 real-world maintenance cases show that G2CP reduces inter-agent communication tokens by 73%, improves task completion accuracy by 34% over free-text baselines, eliminates cascading hallucinations, and produces fully auditable reasoning chains. G2CP represents a fundamental shift from linguistic to structural communication in multi-agent systems, with implications for any domain requiring precise agent coordination. Code, data, and evaluation scripts are publicly available.
Paper Structure (112 sections, 6 theorems, 4 equations, 3 figures, 6 tables, 3 algorithms)

This paper contains 112 sections, 6 theorems, 4 equations, 3 figures, 6 tables, 3 algorithms.

Table of Contents

  1. Introduction
  2. Intuition.
  3. Contributions
  4. Related Work
  5. Agent Communication Languages
  6. Semantic Heterogeneity: FIPA Ontologies vs. G²CP
  7. Multi-Agent LLM Systems
  8. Knowledge Graphs and Agent Systems
  9. The G²CP Protocol
  10. Formal Definition
  11. Intuition.
  12. Operational Semantics
  13. Intuition.
  14. Protocol Properties
  15. Practical implication.
  16. ...and 97 more sections

Key Result

theorem 1

For a fixed graph state $G$ and operation $op$, the result of executing $op$ is deterministic and independent of agent implementation.

Figures (3)

  • Figure 1: G²CP message sequence for a diagnostic query. Natural language (black) is used only at the user boundary; all inter-agent communication (blue) uses G²CP graph operations. Labels: REQ = REQUEST.
  • Figure 2: Task completion accuracy by query category
  • Figure 3: Response time scaling with knowledge graph size. G²CP maintains sub-linear scaling ($O(n^{0.7})$); FTMA scales super-linearly ($O(n^{1.3})$).

Theorems & Definitions (7)

  • definition 1: Graph Operation
  • theorem 1: Determinism
  • theorem 2: Auditability
  • theorem 3: Completeness
  • theorem 4: Traversal Complexity
  • theorem 5: Communication Efficiency
  • theorem 6: Non-Hallucination