start

The start skill acts as the entry point for the Unwind reverse engineering workflow, enabling AI agents to systematically ingest and map an unfamiliar codebase. It is designed for engineers, security auditors, and system architects who need to grasp the high-level structural design of complex projects before performing deep-dive analysis. By analyzing build systems, directory structures, and configuration files, the agent autonomously identifies architectural layers such as database, service, API, and frontend components, producing a structured architecture.md document that serves as the foundation for subsequent modernization or migration planning.

• Automatically detects technology stacks including Node.js, Java, Python, Go, Rust, and .NET based on project configuration files.

• Performs systematic directory mapping to isolate domain models, controllers, service layers, and data access patterns.

• Generates a machine-parseable architecture document providing a unified source of truth for specialized downstream analysis tools.

• Evaluates confidence levels for detected architectural layers to ensure accuracy in complex or legacy environments.

• Identifies cross-cutting concerns like authentication, logging, and error handling patterns spanning the entire application.

• Execute this skill at the very beginning of the reverse engineering process to establish the context for all subsequent tasks.

• Always review the generated docs/unwind/architecture.md to ensure the agent’s detection matches the codebase reality before proceeding to layer-specific analysis.

• Leverages git metadata to create robust, linkable references to source code for all identified components.

• The tool requires read-only access to the repository to safely perform discovery without modifying core application files.

• Use the provided repository information block for seamless integration with IDEs and documentation systems.