Dotnet Service Review

dotnet-service-review

A Claude Code plugin that conducts formal health reviews of .NET microservice repositories using four specialist agents working in parallel.

Usage

Navigate to any .NET repo and run:

/dotnet-service-review:dotnet-service-review

Four agents launch in parallel, analyze the repo, and produce SERVICE-REVIEW.md.

Scoping a Subdirectory

For monorepos or multi-service solutions, scope the review to a specific service:

/dotnet-service-review:dotnet-service-review src/PaymentService

The orchestrator will automatically warn if the target is too large (>5 projects or >500 source files) and ask you to narrow scope.

Run Agents Individually

Each agent is also available standalone:

/dotnet-service-review:architect       # Architecture & data flow diagram
/dotnet-service-review:code-review     # Code quality, .NET practices, security
/dotnet-service-review:test-analysis   # Build, tests, testing trophy, maintainability

No configuration required. The repo you're in IS the service.

Providing Institutional Context

Create a SERVICE-REVIEW-CONTEXT.md in the repo root to give the agents domain knowledge they can't discover from code alone:

# Service Review Context

## Domain
This service is part of the Batch Processing domain. Related repos: core, listener-rabbitmq.

## Known Issues
- The GenerateACHFile class uses virtual methods (WriteBatchHeader, WriteDetail, etc.)
  to support custom ACH DLLs via inheritance. Refactoring must preserve this contract.
- CheckMarx has flagged SQL injection in GetRecordsCount repeatedly; prior fixes
  addressed symptoms, not root cause.

## Team
Single active contributor. Bus factor = 1. Onboarding documentation is sparse.

## Architecture Notes
- Shares dbo.PAYMENTS table with core — no domain API boundary exists.
- OFAC screening is documented in wiki as happening here, but actually lives in core.

This file is optional. When present, its contents are passed to all four agents to inform their analysis. It's particularly valuable for:

• Context that lives in people's heads, not in code
• Known issues or technical debt that should be verified, not rediscovered
• Cross-repo relationships that aren't visible from this repo alone
• Historical context (past security scans, incident patterns, failed refactors)

How It Works

The orchestrator launches four specialist agents in parallel:

Agent	Analyzes	Scores	Key MCP Tool
dotnet-architect	Architecture, data flow, integration points, wiki docs	—	GitNexus
dotnet-code-reviewer	SOLID, .NET practices, security, vulnerabilities	Code Quality, .NET Practices, Security	RoslynMCP
dotnet-test-analyst	Build, tests, dependencies, documentation	Testing, Maintainability	Bash
dotnet-modernization-analyst	Framework targeting, API compatibility, migration complexity	Modernization Readiness	RoslynMCP

The orchestrator collects their outputs, computes the overall health score, synthesizes recommendations, and writes the final report.

Tool Availability Reporting

The code reviewer performs a mandatory RoslynMCP probe at the start of analysis. If RoslynMCP is unavailable, the review proceeds with text-based analysis and the final report includes a warning indicating reduced analysis depth. Tool availability for all agents is surfaced in the report's Additional Notes section.

MCP Servers

Works with zero MCP servers. Significantly deeper with them.

MCP Server	Used By	What It Adds
GitNexus	architect	Cross-repo dependency mapping, blast radius analysis
RoslynMCP	code-reviewer, modernization-analyst	Cyclomatic complexity, compiler diagnostics, symbol analysis, framework compatibility
Documentation (Confluence, Notion)	architect	Wiki context for architectural decisions

Report Includes

• Overall health score with 5-category breakdown
• Architecture overview with Mermaid data flow diagram
• Key integration points table
• Build & test results
• Testing trophy (current vs. recommended)
• Modernization readiness assessment with migration complexity
• Detailed findings per category with file:line references
• Tool availability table showing analysis depth
• Top 3 prioritized recommendations

Scoring

Traffic-light (Green/Orange/Red) across 5 categories:

GREEN  = 0 Red AND <= 2 Orange
ORANGE = 1-2 Red OR 3+ Orange
RED    = 3+ Red OR any Critical failure

Modernization Readiness is scored separately and does not affect the overall health score.

Plugin Structure

dotnet-service-review/
├── .claude-plugin/
│   └── plugin.json
├── agents/
│   ├── dotnet-service-review.md      # Orchestrator
│   ├── dotnet-architect.md           # Architecture specialist
│   ├── dotnet-code-reviewer.md       # Code quality specialist
│   ├── dotnet-test-analyst.md        # Test & build specialist
│   └── dotnet-modernization-analyst.md  # Modernization specialist
├── skills/
│   ├── dotnet-service-review/
│   │   ├── SKILL.md                  # Full review (orchestrated)
│   │   └── knowledge/               # 8 shared knowledge files
│   ├── architect/
│   │   └── SKILL.md                  # Standalone architecture analysis
│   ├── code-review/
│   │   └── SKILL.md                  # Standalone code review
│   └── test-analysis/
│       └── SKILL.md                  # Standalone test analysis
└── README.md

License

MIT