nx-mcp

Exposes Nx task execution (build, test, lint, serve) as MCP tools that AI clients can invoke directly. Implements the Model Context Protocol server specification to translate natural language task requests into Nx CLI commands, handling task graph resolution, dependency ordering, and parallel execution configuration. Routes execution through Nx's task scheduler rather than shelling out, enabling real-time progress streaming and structured result parsing.

Provides MCP tools that introspect the Nx workspace to enumerate all projects, their targets (build, test, lint, etc.), dependencies, and configuration. Parses nx.json, project.json files, and plugin metadata to build a queryable index of available tasks and their parameters. Returns structured metadata (project graph, target configurations, affected projects) that AI clients can use to understand workspace structure without manual exploration.

Provides MCP tools for git operations within the Nx workspace context, including file change detection, commit history analysis, and branch management. Integrates with Nx's affected detection to correlate git changes with project impacts. Enables AI clients to understand code history and make informed decisions about which projects to rebuild or test.

Implements Nx's affected command as an MCP tool, analyzing file changes (via git diff or provided file list) to determine which projects in the monorepo are impacted. Uses Nx's dependency graph and file-to-project mapping to compute the minimal set of projects that need re-testing or rebuilding. Returns structured output (affected projects, their targets, and change scope) that AI agents can use to optimize CI/CD workflows.

Exposes Nx generators (schematics-based code generation) as MCP tools, allowing AI clients to invoke generators for creating components, services, libraries, and other boilerplate. Parses generator schemas to expose configurable options as MCP tool parameters, handles generator execution with proper file I/O and git integration, and returns structured output (generated files, paths, next steps). Supports both built-in Nx generators and custom workspace generators.

Provides MCP tools to query and analyze Nx's project dependency graph, including transitive dependencies, circular dependency detection, and dependency path analysis. Returns graph data in structured formats (adjacency lists, edge lists) suitable for visualization or algorithmic analysis. Enables AI agents to understand project relationships, identify tightly-coupled modules, and suggest refactoring opportunities.

Exposes Nx lint targets (ESLint, TSLint, custom linters) as MCP tools, allowing AI clients to run linting rules and retrieve structured violation reports. Parses linter output (JSON format) to provide machine-readable results including file paths, line numbers, rule names, and suggested fixes. Integrates with Nx's caching to avoid re-linting unchanged files, and supports auto-fix capabilities where available.

Exposes Nx test targets (Jest, Vitest, Cypress, etc.) as MCP tools, enabling AI clients to run tests and retrieve structured results. Parses test output (JSON format) to provide machine-readable results including test names, pass/fail status, execution time, and error messages. Integrates with Nx's caching to skip re-running passing tests, and supports filtering by test name or file path.

Exposes Nx build targets as MCP tools, enabling AI clients to trigger builds and retrieve information about generated artifacts. Parses build output to provide structured metadata (output paths, bundle sizes, asset lists, build time). Integrates with Nx's caching to skip rebuilds of unchanged code, and supports multiple build configurations (dev, prod, etc.).

Provides MCP tools to read and modify Nx workspace configuration (nx.json, project.json files). Enables AI clients to query current settings, update task configurations, modify project metadata, and manage workspace-level settings. Handles JSON parsing and validation, and supports atomic updates with rollback capability.

Exposes Nx's plugin system as MCP tools, allowing AI clients to discover installed plugins, their capabilities, and available generators/executors. Queries plugin metadata to provide information about plugin-provided targets, generators, and configuration options. Supports plugin installation and configuration through MCP.