feat: set default model to sonnet with opus tiering

rules/claude-code-builtins.md

@@ -0,0 +1,57 @@
+- Built-in slash commands and CLI flags are handled by the Claude Code harness, not by Claude. They are invisible during a session — do not attempt to run them via Bash or simulate their behavior.
+- When the user needs a built-in action, tell them the command to run (e.g., "Run `/model opus`") rather than trying to execute it yourself.
+- New builtins appear in release notes — flag them during `/cc-release-review`.
+
+## Slash commands (run inside session with `/`)
+
+### Session
+
+`/clear` `/resume [session]` `/branch [name]` `/rename [name]` `/exit` `/btw <question>`
+
+### Model & effort
+
+`/model [model]` `/effort [level]` `/fast [on|off]`
+
+### Configuration
+
+`/config` `/permissions` `/keybindings` `/terminal-setup` `/statusline` `/sandbox` `/login` `/logout` `/privacy-settings` `/memory` `/hooks` `/extra-usage`
+
+### Information
+
+`/help` `/status` `/cost` `/context` `/stats` `/insights` `/release-notes` `/skills` `/doctor` `/usage`
+
+### Code & project
+
+`/diff` `/copy [N]` `/export [filename]` `/rewind` `/security-review` `/plan [description]` `/ultraplan <prompt>` `/add-dir <path>`
+
+### Tools & integrations
+
+`/mcp` `/plugin` `/ide` `/chrome` `/reload-plugins`
+
+### Remote & scheduling
+
+`/schedule` `/desktop` `/remote-control` `/remote-env`
+
+### GitHub & external
+
+`/install-github-app` `/install-slack-app`
+
+### Tasks
+
+`/tasks`
+
+### Bundled skills (not user-defined)
+
+`/simplify` `/batch` `/debug` `/loop`
+
+## CLI commands (run from terminal)
+
+`claude` `claude -p "query"` `claude -c` `claude -r "<session>"` `claude -w [name]` `claude --remote` `claude --teleport` `claude update` `claude auth login|logout|status` `claude agents` `claude mcp` `claude plugin` `claude auto-mode defaults|config`
+
+## Key CLI flags
+
+`--model` `--effort` `--permission-mode` `--tools` `--disallowedTools` `--add-dir` `--agent` `--system-prompt` `--max-budget-usd` `--max-turns` `--output-format` `--settings` `--mcp-config`
+
+## Keyboard shortcuts (session)
+
+`Shift+Tab` / `Alt+M` — cycle permission modes · `Option+P` — switch model · `Option+T` — toggle thinking · `Option+O` — toggle fast mode · `Ctrl+B` — background tasks · `Ctrl+G` — open in editor · `@` — file autocomplete · `!` — bash mode

rules/model-selection.md

@@ -0,0 +1,3 @@
+- When about to invoke a superpowers skill that requires deep reasoning — specifically `brainstorming`, `systematic-debugging`, `writing-plans`, or `subagent-driven-development` — check the current model from the system prompt ("You are powered by the model named..."). If not on Opus, suggest the user run `/model opus` before proceeding.
+- Keep the nudge brief: "This skill works best on Opus — run `/model opus` to switch."
+- Don't block on it — if the user proceeds without switching, continue normally.

settings.json

@@ -1,5 +1,8 @@
 {
   "$schema": "https://json.schemastore.org/claude-code-settings.json",
+  "model": "sonnet",
+  "spinnerTipsEnabled": false,
+  "prefersReducedMotion": true,
   "permissions": {
     "allow": [
       "Edit",
@@ -39,9 +42,25 @@
       "Bash(gofumpt:*)",
       "Bash(golangci-lint:*)",
       "Bash(exiftool:*)",
+      "Bash(awk:*)",
+      "Bash(cat:*)",
+      "Bash(cut:*)",
+      "Bash(diff:*)",
+      "Bash(echo:*)",
+      "Bash(find:*)",
+      "Bash(grep:*)",
+      "Bash(head:*)",
+      "Bash(obsidian:*)",
       "Bash(jq:*)",
+      "Bash(printf:*)",
+      "Bash(sed:*)",
+      "Bash(sort:*)",
+      "Bash(tail:*)",
+      "Bash(tr:*)",
       "Bash(tsc:*)",
+      "Bash(uniq:*)",
       "Bash(wc:*)",
+      "Bash(xargs:*)",
       "WebFetch(domain:code.claude.com)",
       "WebFetch(domain:docs.anthropic.com)",
       "WebFetch(domain:github.com)",

skills/cc-release-review/SKILL.md

@@ -61,6 +61,7 @@ For each item in the release notes, classify it and check relevance:
 
 - Does this introduce a setting the user might want? Check if it's already configured.
 - Does it add a new hook event type? Check if the user's log-event wiring or other hooks should cover it.
+- Does it add a new slash command or CLI flag? Check if `rules/claude-code-builtins.md` needs updating.
 - Does it change how an existing feature works in a way that affects current config?
 
 #### Removed features

skills/fix-issue/SKILL.md

@@ -1,5 +1,6 @@
 ---
 argument-hint: "[issue-number]"
+model: opus
 allowed-tools: Read, Edit, Write, Bash, Grep, Glob
 description: Plans, implements, reviews, and ships a fix for a GitHub issue. Use when fixing an issue, resolving a bug report, implementing a feature request, or closing a ticket. Handles fork workflows, upstream detection, branch creation, and PR submission.
 ---

skills/local-issues/SKILL.md

@@ -1,5 +1,6 @@
 ---
 argument-hint: "[path or scope]"
+model: opus
 allowed-tools: Read, Grep, Glob, Bash, Write
 description: Reviews a codebase for bugs, design issues, and code cleanliness problems with specific file paths and line numbers. Use when auditing code quality, finding bugs, doing a code review, finding problems, or reviewing a project for issues. Creates issue files in `.issues/` directory.
 ---

skills/refactor-clean/SKILL.md

@@ -1,4 +1,5 @@
 ---
+model: opus
 description: Structured refactoring with smell detection, severity classification, and before/after metrics. Use when code needs deep structural analysis — decomposing large classes, resolving SOLID violations, eliminating duplication across modules, or reducing cyclomatic complexity. Presents a prioritized plan for approval before making changes. Not for lightweight post-edit polish (the simplify agent handles that automatically).
 ---
 

skills/review-fix/SKILL.md

@@ -1,4 +1,5 @@
 ---
+model: opus
 description: "Runs an iterative review-fix loop on code changes. Use when reviewing a diff and fixing findings in place — presents findings by severity, lets you choose which to fix, applies fixes, and re-reviews. Max 3 iterations."
 allowed-tools: Read, Edit, Write, Bash, Grep, Glob, Agent, AskUserQuestion
 ---

skills/scientific-problem-selection/SKILL.md

@@ -1,4 +1,5 @@
 ---
+model: opus
 description: "Guides scientists through systematic research problem selection using a 9-skill framework based on Fischbach & Walsh (Cell, 2024). Use when pitching a new research idea, troubleshooting a stuck project, evaluating project risks, planning research strategy, or choosing what problem to work on. Covers ideation, risk assessment, optimization, parameter strategy, decision trees, adversity planning, and synthesis."
 allowed-tools: Read, AskUserQuestion, Bash, WebSearch
 context: fork