Consider removing CLAUDE.md and amazonq.md - redundant with ai-provider-agnostic setup?

## Question
Do we still need `CLAUDE.md` and `amazonq.md` files given our AI provider agnostic setup?

## Current State
We've implemented a unified knowledge base system:
- `knowledge/` directory contains all principles and procedures
- Amazon Q uses symlinked `~/.amazonq/rules/` → `knowledge/`
- Claude Code uses generated `CLAUDE.local.md` files that pull from `knowledge/`
- All providers reference the same single source of truth

## What CLAUDE.md Currently Contains
```markdown
# AI Provider Context

This repository uses a unified knowledge base for all AI providers.

## Global Context
See `knowledge/` directory for principles and procedures that apply across all projects.

## Repository-Specific Context
- Repository: atxtechbro/dotfiles
- Primary workflow: GitHub Pull Requests
- See README.md for repository-specific principles

## Critical: Non-Interactive Execution Only
See `knowledge/procedures/non-interactive-execution.md` for the fundamental principle.

Key insight: You (Claude) are BY DEFINITION already in an interactive session. You literally CANNOT provide interactive input to any command. This includes `claude` commands which must use `-p` flag.

## For AI Providers
- **Amazon Q**: Uses symlinked `~/.amazonq/rules/` → `knowledge/`
- **Claude Code**: Uses generated `CLAUDE.local.md` files
- **Others**: See `docs/ai-provider-agnostic-context.md` for integration patterns

All providers should reference the same `knowledge/` directory as the single source of truth.
```

## Analysis
Most of this content either:
1. Duplicates information already in `docs/ai-provider-agnostic-context.md`
2. Points to the `knowledge/` directory (which providers already access)
3. Contains provider-specific notes that could live in `knowledge/procedures/`

The only unique content is the repository-specific context, which could move to:
- `knowledge/repository-context.md`
- Or directly into the relevant procedures

## Proposal
1. Move any unique/valuable content to appropriate locations in `knowledge/`
2. Delete `CLAUDE.md` and `amazonq.md`
3. Update `setup.sh` if needed to stop creating these files
4. Ensure all AI providers still work correctly with just the `knowledge/` directory

## Questions
- Is there any provider-specific content that MUST live outside `knowledge/`?
- Do the generated `CLAUDE.local.md` files already include everything needed?
- Would removing these files break any existing workflows?

**Principle**: `subtraction-creates-value` - removing redundant files reduces confusion

Compare CLAUDE.md and amazonq.md with docs/ai-provider-agnostic-context.md, the knowledge/ directory, and setup.sh to identify unique content and creation logic. Check whether generated CLAUDE.local.md files and the Amazon Q symlink still provide the required context, then verify that removing the redundant files does not break provider workflows.

Written by the indexing model from the issue text.