Add comprehensive concept documentation for ra

[chinmaymk]

Summary

This PR adds a complete set of concept documentation pages that explain the core ideas and components of ra. These pages serve as the conceptual foundation for users to understand how ra works before diving into detailed references.

Key Changes

• Added 7 new concept documentation pages:

  • agent-loop.md — Explains the core agent loop cycle, middleware hooks, and configuration options
  • providers.md — Documents available LLM providers, switching between them, and credential management
  • tools.md — Describes built-in tools, MCP integration, permissions, and tool execution flow
  • middleware.md — Covers the middleware system, all 9 hooks, and how to write custom middleware
  • skills.md — Explains skill structure, activation, scripts, references, and skill directories
  • recipes.md — Documents recipe packaging, installation, config layering, and built-in recipes
  • configuration.md — Details the layered config system, environment variable interpolation, and minimal/full config examples
  • sessions.md — Covers session persistence, resuming, scoping, auto-pruning, and the memory system

• Updated navigation structure:

  • Reorganized sidebar to separate "Concepts" (new) from "Core" sections
  • Added all 8 new concept pages to the Concepts navigation menu
  • Moved "The Agent Loop" reference to Core section (keeping concept version in Concepts)

• Added onboarding link:

  • Added a "New to ra? Start with Concepts" link to the homepage to guide new users

Implementation Details

• All documentation follows consistent structure with clear headings, code examples, and tables
• Concept pages focus on "what" and "why" rather than exhaustive API details
• Cross-references between pages use relative links (e.g., /concepts/agent-loop)
• Examples use realistic YAML configurations and TypeScript code
• Each page includes a reference link to more detailed documentation where applicable

https://claude.ai/code/session_01Wkd4XjYwEGSCNwzcfkthDJ