Prevent cross-repo architectural breakage before merge.

Enforce architectural policies in CI, evaluate blast radius, and give AI agents live system context.

CodeRadius ingests your repositories and constructs a persistent, queryable graph of your entire architecture — services, APIs, databases, message queues, teams, and the relationships between them. Then it does three things:

1. Blocks architectural violations in CI — declarative policy rules evaluated against the live graph, not file-level lint.
2. Predicts blast radius before merge — in-memory topological diff that finds downstream consumers of your change in under 10 seconds.
3. Gives AI agents architectural context — a native MCP server that lets Cursor, Claude, or Windsurf query the graph before writing code.

The Problem

AI coding agents and large engineering teams share the same blind spot: they optimize locally and break globally.

An agent renames a JSON field in a message payload — unaware that a consumer three teams away parses it. An engineer refactors a database table — unaware that six services read from it via a shared query. A new team scaffolds a service without CI/CD, without ownership metadata, without any of the standards the rest of the organization follows.

These problems are not solvable with file-level linters, code search, or Slack threads. They require a cross-repo topological model — a live graph of how services, APIs, databases, and teams are connected.

CodeRadius builds that graph and makes it actionable.

Core Capabilities

1. Governance — Policy Enforcement Across the Fleet

CodeRadius includes a Governance-as-Code engine that evaluates declarative YAML rules against the architecture graph. Each rule is a Cypher query that identifies non-compliant entities — repositories, services, or packages — and surfaces them as structured violations.

These are not lint rules. They are topology-level checks that cross-reference service exposure, team ownership, dependency health, CI/CD configuration, and database patterns across the entire graph.

Examples:

• Service exposes a public API but depends on deprecated packages
• Repository has no CI/CD pipeline configuration
• Multiple services access the same database table without a data contract
• Service has no declared team ownership

→ Governance & Golden Path

2. Impact Evaluation — Blast Radius Before Merge

cr blast runs an in-memory topological diff of your code changes against the live graph. It detects breaking changes (deleted or modified relationships to shared resources), finds all downstream consumers, and returns a structured finding report.

• Runs locally in 2–5 seconds
• Returns exit code 1 on breaking changes — blocks CI
• Supports --advisory mode for gradual rollout
• Outputs Markdown for PR comment injection (GitHub Actions, GitLab CI, Bitbucket)

Think of it as terraform plan for your architecture.

→ Impact Evaluation

3. MCP Context — Source of Truth for AI Agents

CodeRadius ships a native Model Context Protocol (MCP) server. When connected to your IDE, AI agents can query the architectural graph in real time:

• Before changing an API: "Who consumes this endpoint?"
• Before renaming a field: "What's the exact data contract?"
• Before proposing a refactor: "What's the blast radius of this change?"

This is what separates CodeRadius from pure code search. The agent doesn't just see files — it sees the cross-repo topology and can reason about downstream impact.

→ MCP Server Reference

Additional Capabilities

Beyond the core workflow, CodeRadius provides:

• Architecture Dashboard — A self-contained HTML report with service inventory, dependency graphs, SPOF analysis, and governance violations.
• System Registry — Auto-generated service catalog of every repository, service, and team.
• SPOFs & Data Gravity — Identifies shared databases, service bottlenecks, and concentration risks ranked by a 0–100 SPOF score.
• Agentic Context Radar — Maps AI tooling adoption across the fleet: maturity levels, capabilities catalog, context gaps, and team coverage metrics.

Who It Is For

Quick Start

Prerequisites

• Node.js ≥ 22 or Bun
• Docker (for the Memgraph graph database)
• LLM API Key (Google Vertex AI, OpenAI, or Anthropic)

curl -sSL https://cdn.coderadius.ai/install.sh | bash

cr init

cr up

cr analyze code ./path/to/your/services/*

# Evaluate blast radius of your current changes
cr blast --base main --head HEAD

# Open the MCP server for your IDE
cr mcp start

# Generate the Architecture Dashboard
cr ui

Supported Languages and Patterns

CodeRadius ships with native support for TypeScript, PHP, Python, and Go — including framework-specific analysis for NestJS, Symfony, FastAPI, Express, Laravel, Gin, and more. It auto-discovers REST APIs, OpenAPI specs, GraphQL resolvers, database connections, and message broker topologies.

→ Full compatibility matrix

Next Steps

• Use Cases — Three concrete scenarios where CodeRadius prevents real production incidents.
• Governance & Golden Path — Define and enforce architectural standards across the fleet.
• Impact Evaluation — Predict the blast radius of code changes before you commit.
• MCP Server — Connect your AI agents to the architecture graph.