LLM Flex 0.1.0

First public alpha. A macOS menu-bar app for switching the model/provider used
by Claude Code and OpenAI Codex without hand-editing config files.

Added

• Profiles — define a provider, base URL, and model; switch between them
  from the menu bar. API keys are stored in the macOS Keychain, never in plain
  config or UserDefaults.
• Claude Code target — writes ~/.claude/settings.json (model + auth env),
  deep-merging so your other settings (permissions, hooks) are preserved.
• Codex target — writes a managed block in ~/.codex/config.toml, updates
  ~/.codex/auth.json, and sets OPENAI_API_KEY in the launchd environment.
  Blocks switching while Codex is signed in via ChatGPT to avoid clobbering.
• Providers — OpenAI, OpenRouter, LM Studio, Anthropic, Gemini, Ollama,
  Opencode Go, plus OpenAI-compatible and custom endpoints.
• Per-target compatibility chips — flags provider/target combos that won't
  work without a translating proxy (e.g. OpenAI ↔ Claude Code).
• Curated model catalog with a Quick Pick picker in the profile editor.
• Test Connection — validates a profile against the provider using its
  real auth scheme and surfaces the provider's error message on failure.
• Snapshot & restore — captures your original config files on first touch
  so changes can be fully rolled back.
• DMG packaging (scripts/package-dmg.sh) and tag-triggered GitHub Releases.