The Problem with System Documentation

Even the most sophisticated architectures fail when their logic is trapped in static PDFs or scattered Markdown files. Humans struggle to navigate the complexity, and AI agents lack the structured context needed to provide meaningful assistance. Standard prompting isn't enough to bridge the gap between "code on disk" and "architectural intent."

What it does

This skill provides a rigorous framework to transform complex systems into agent-readable knowledge bases. It goes beyond simple writing by designing MCP (Model Context Protocol) interfaces that allow AI agents to programmatically query your architecture, decision records, and system state.

• Audit & Assessment: Evaluates current docs for agent-blind spots and structural gaps.
• Architecture Design: Defines a layered documentation strategy optimized for both human reading and LLM token-efficiency.
• MCP Integration: Maps your system's resources and tools to specific agent-facing interfaces.
• Evolution Workflows: Establishes "living doc" patterns to ensure documentation never drifts from implementation.

Why use this skill?

Generic AI prompts often guess at system intent, leading to hallucinations or "lazy" solutions. This skill structures your project so that any agent (like Claude Code or Cursor) can instantly understand the why behind your code, resulting in higher-quality contributions and faster onboarding for new developers.