Architecture Intelligence Platform

•
10
•

Provides trusted architecture context for AI agents by building an evidence-backed graph from declared specifications and runtime telemetry.

The Architecture Intelligence Platform (AIP) empowers AI agents with a reliable understanding of your software architecture. It constructs an evidence-backed model by integrating declared specifications like OpenAPI, AsyncAPI, and architecture manifests with real-world runtime telemetry from OpenTelemetry. This enables agents to distinguish between declared, observed, and reconciled architectural facts, improving the accuracy of their analyses and changes. AIP exposes this validated architecture through read-only Model Context Protocol (MCP) tools, offering insights into dependencies, architecture drift, and the provenance of every architectural claim, all bound to a specific graph snapshot.

주요 기능

01Exposes validated architecture via three read-only Model Context Protocol (MCP) tools: get_service_dependencies, get_architecture_drift, and get_evidence
020 GitHub stars
03Generates evidence-backed architecture graphs from OpenAPI, AsyncAPI, and OpenTelemetry
04Reconciles declared specifications with observed runtime behavior, identifying confirmed, observed-only, and not-observed-in-window dependencies
05Provides deterministic dependency analyses, architecture blast radius calculations, and telemetry coverage qualification
06Uses Neo4j as its architecture knowledge graph, ensuring atomic, evidence-linked fact storage

사용 사례

01Discover undocumented service dependencies by analyzing real traffic patterns
02Provide reliable, citeable architectural context to AI coding agents for more accurate and safer modifications
03Identify architecture drift by comparing declared specifications against actual runtime behavior