advance-minimax-m2-cursor-rules

Generates structured clarification prompts before code generation by decomposing user intent into explicit requirements, constraints, and context. Uses a multi-turn prompt engineering pattern that forces the AI to ask disambiguating questions about scope, dependencies, error handling, and testing before writing code, reducing hallucination and scope creep in generated artifacts.

Leverages MiniMax M2's native interleaved thinking capability to expose intermediate reasoning steps during code generation and analysis. The system chains thinking tokens with code generation, allowing the AI to reason about architectural decisions, trade-offs, and implementation details before committing to code, with reasoning visible to the developer for transparency and debugging.

Implements a schema-based function registry that maps user intents to executable tools (file operations, API calls, test execution, deployment) with native bindings for MiniMax M2's function-calling API. The system manages tool sequencing, error handling, and state propagation across multi-step workflows, enabling the AI to autonomously orchestrate complex coding tasks like testing, linting, and deployment without manual intervention.

Maintains an indexed representation of the developer's codebase within the MCP server, enabling the AI to retrieve relevant code context, dependencies, and patterns without sending the entire codebase to the LLM on each request. Uses semantic understanding of code structure to surface related files, function signatures, and architectural patterns that inform code generation decisions.

Generates code with built-in error handling patterns, type safety, and test coverage by composing generation prompts with explicit requirements for exception handling, input validation, and unit test generation. The system uses MiniMax M2's reasoning to consider edge cases and failure modes before generating code, then optionally executes generated tests via tool orchestration to validate correctness.

Maintains conversation state and reasoning context across multiple turns within a Cursor session, allowing the AI to build on previous decisions, refine code iteratively, and track architectural decisions across a coding session. Uses MCP server-side state management to persist context between requests, enabling the AI to reference earlier reasoning and avoid redundant analysis.

Provides a framework for defining and customizing Cursor Rules (system prompts for Cursor IDE) using template variables, conditional logic, and modular rule composition. Allows developers to create reusable rule sets tailored to specific projects, languages, or coding standards, with MiniMax M2 optimizations baked into the rule templates.

Encodes language and framework-specific best practices, idioms, and patterns into the code generation pipeline, enabling the AI to generate code that follows language conventions, uses idiomatic patterns, and respects framework constraints. Includes specialized handling for type systems, async patterns, dependency management, and framework-specific APIs.