roslyn-mcp

Solution health check. Use when: analyzing a C# solution or project, checking build health, finding diagnostics, assessing code quality, or getting an overview of a .sln/.csproj. Takes a path to a solution or project file as input.

Layering and Dependency-Inversion-Principle audit of a C# solution. Use when: auditing layering, detecting cycles, checking DIP compliance, or finding cross-layer leaks. Takes an optional project or namespace filter as input.

Roslyn code actions (fixes and refactorings). Use when: applying IDE-style quick fixes or refactorings at a position or selection, including introduce parameter and inline temporary variable. Takes file path and line/column (and optional selection end) as input.

Complexity hotspot analysis. Use when: finding complex methods, identifying god classes, measuring maintainability, or planning refactoring priorities in a C# solution. Optionally takes a project name.

Dead code detection and cleanup. Use when: finding unused symbols, removing dead code, cleaning up unreferenced private/internal members, or auditing a C# project for code that can be safely deleted. Optionally takes a project name.

Audit dependency-injection registrations in a .NET solution. Use when: auditing DI registrations, finding missing bindings, composition-root review, or detecting lifetime mismatches (Singleton depending on Scoped, etc.). Produces a layered report of every registered service, missing interfaces, unresolved implementations, and DI anti-patterns.

C# XML documentation generator. Use when: adding XML doc comments, documenting public APIs, fixing CS1591 warnings, or improving documentation quality in *.cs files. Takes a file path or type name as input.

Repo-wide classification of exception handling patterns. Use when: auditing exception handling, finding swallowed exceptions, classifying catch patterns, finding overly-broad catches, or mapping unhandled-at-boundary throw sites in a C# solution. Optionally takes an exception type name to filter the audit.

Diagnostic explainer and fixer. Use when: understanding a C# compiler error or warning, finding fixes for diagnostics, batch-fixing all instances of a diagnostic, or troubleshooting build failures. Takes a diagnostic ID (e.g., CS0246) or file:line as input.

Extract method refactoring. Use when: extracting a block of statements into a new method, reducing method complexity, or breaking up long methods. Describe the code region and target method name as input.

Whole-solution formatting compliance pass. Use when: enforcing whole-solution formatting, running `dotnet format` equivalents, bulk-organizing usings, or performing an editorconfig compliance sweep. Optionally takes a project name or file glob to narrow scope.

Batch-scaffold test stubs for untested public APIs. Use when: batch-scaffolding tests for untested public APIs, generating test stubs, or bootstrapping a test project. Takes a top-N count, a project name, or a type name.

Pre-change blast-radius report for a C# symbol. Use when: planning a rename, signature change, or deletion and you need to know how far the change reaches before touching code. Takes a symbol name (and optional change type) as input. Produces a ranked safe-to-change verdict with direct references, semantic consumers, polymorphic impact, and mutation sites.

Walk the inheritance and member-override graph for a type or member. Use when: exploring type hierarchies, finding overrides, finding implementations, understanding polymorphic dispatch, auditing interfaces. Takes a type or member name as input.

NuGet package migration. Use when: replacing one NuGet package with another across a solution, upgrading packages, or migrating from deprecated packages. Takes old package name, new package name, and new version as input.

Staged codebase modernization. Use when: systematically modernizing a C# codebase toward a stated goal, migrating between patterns/attributes/APIs at scale, or adopting newer language features (file-scoped namespaces, primary constructors, collection expressions) across many files. Takes a natural-language modernization goal as input.

Pre-NuGet-publish readiness check. Use when: preparing to publish to NuGet, checking release readiness for a .NET library, or validating package metadata, build, tests, and vulnerabilities before a release cut. Optionally takes a project name to narrow the scope.

MSBuild and project-file inspection. Use when: debugging TargetFramework, OutputPath, package references, item includes, or evaluated MSBuild properties for a .csproj.

Guided refactor → preview → apply → validate loop using v1.17/v1.18 primitives. Use when: implementing a non-trivial refactor that benefits from explicit staging through preview, apply-with-verify, and the validate_workspace bundle. Pairs well with the refactor skill for upstream symbol-level work.

Guided semantic refactoring. Use when: renaming symbols, extracting interfaces, extracting types, moving types between files or projects, splitting classes, or performing bulk type replacements in C# code. Describe the desired refactoring as input.

Semantic code review. Use when: reviewing C# code quality, finding issues before a PR, auditing a file or project for problems, or doing a comprehensive quality check. Optionally takes a file path or project name as input.

Security audit for C# solutions. Use when: auditing for vulnerabilities, checking NuGet packages for CVEs, reviewing security diagnostics, finding reflection usage, auditing DI registrations, or doing an OWASP-style security review. Optionally takes a project name.

Natural-language code search across a C# codebase. Use when: looking for a class, method, or pattern described in plain English (e.g. 'the class that handles payment refunds', 'methods that retry on failure'), locating code by behavior rather than exact name, or orienting in an unfamiliar solution. Takes a natural-language description as input.

Session history and undo. Use when: the user wants to see what Roslyn MCP changed, undo the last apply, reload the workspace from disk, or recover from a bad refactoring.

Quick C# feedback without loading a solution. Use when: validating syntax, semantics, or running a small script; prototyping an expression; or explaining an isolated code fragment.

Test coverage analysis. Use when: checking test coverage, finding untested code, identifying gaps in test suites, scaffolding new tests, or auditing which public APIs have tests. Optionally takes a project name.

Test discovery and failure triage. Use when: CI is red, tests fail locally, or you need to find and run the right tests after a change. Optionally takes project name or filter.

Walk a value, parameter, or symbol through control flow, data flow, and exception flow paths. Use when: tracing a value through control/data/exception flow, understanding how a parameter propagates, finding where an exception is caught, or answering 'where can this value go?' / 'what paths does control take?' / 'if this throws, who catches it?'.

Update the Roslyn MCP plugin. Use when: server_info shows an update is available, the user wants to update to the latest version, or the plugin reports an older version than NuGet. Handles both the global tool binary (Layer 1) and the Claude Code plugin metadata (Layer 2).

Multi-project version bump across a .NET solution. Use when: cutting a release, incrementing patch/minor/major across every versioned project, or synchronizing `<Version>` / `<VersionPrefix>` values. Takes a bump type (patch, minor, or major) as input. Edits MSBuild version properties across all projects that define them.

One-shot status report on the running Roslyn MCP server and any loaded workspaces. Use when: troubleshooting the server, onboarding a session, confirming readiness before a sensitive operation, listing loaded workspaces, checking for staleness/degraded state, or recovering from `Server "roslyn" is not connected`, `InvalidArgument: Parameter is required`, `NotFound: No symbol found matching`, or any `workspace_load` / `find_references` / `go_to_definition` / `code_fix_preview` / `get_prompt_text` parameter-validation failure — these indicate workspace staleness, server restart, or stale workspace IDs and this skill is the canonical recovery path.

4 hooks across 2 events