@torka/claude-qol 0.1.0

package/.claude-plugin/plugin.json

@@ -0,0 +1,6 @@
+{
+  "name": "@torka/claude-qol",
+  "version": "0.1.0",
+  "description": "Claude Code quality-of-life improvements: auto-approve hooks and context monitoring",
+  "type": "hooks"
+}

package/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2024 Varun Torka
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

package/README.md

@@ -0,0 +1,266 @@
+# @torka/claude-qol
+
+Claude Code quality-of-life improvements: auto-approve hooks, context monitoring, and workflow optimization.
+
+## Features
+
+| Feature | Description |
+|---------|-------------|
+| **Auto-Approve Hook** | Intelligent command auto-approval for solo dev workflows |
+| **Context Monitor** | Real-time context usage status line with visual indicators |
+| **Auto-Format Hook** | Run linters/formatters automatically after file edits |
+| **Completion Notifications** | Desktop notifications + sound when tasks complete |
+| **Optimize Command** | Analyze auto-approve decisions and refine rules |
+
+## Installation
+
+### Project-level (recommended)
+
+```bash
+npm install --save-dev @torka/claude-qol
+```
+
+### Global
+
+```bash
+npm install -g @torka/claude-qol
+```
+
+The installer automatically copies files to your `.claude/` directory:
+- **Project-level**: Files go to `<project>/.claude/`
+- **Global**: Files go to `~/.claude/`
+
+## Post-Installation Setup
+
+### 1. Configure Auto-Approve Hook (recommended)
+
+Add to your `.claude/settings.local.json`:
+
+```json
+{
+  "hooks": {
+    "PreToolUse": [
+      {
+        "matcher": "Bash|Read|Grep|Glob|Write|Edit|MultiEdit",
+        "hooks": [
+          {
+            "type": "command",
+            "command": "python3 .claude/scripts/auto_approve_safe.py"
+          }
+        ]
+      }
+    ]
+  }
+}
+```
+
+### 2. Configure Status Line (optional)
+
+Add to your `.claude/settings.local.json`:
+
+```json
+{
+  "statusLine": {
+    "type": "command",
+    "command": "python3 .claude/scripts/context-monitor.py"
+  }
+}
+```
+
+This shows a real-time context usage bar with percentage and warnings.
+
+### 3. Auto-Format on Edit (optional)
+
+Auto-run linters and formatters after file edits:
+
+```json
+{
+  "hooks": {
+    "PostToolUse": [
+      {
+        "matcher": "Edit|MultiEdit",
+        "hooks": [
+          {
+            "type": "command",
+            "command": "if [[ \"$CLAUDE_TOOL_FILE_PATH\" == *.js || \"$CLAUDE_TOOL_FILE_PATH\" == *.ts || \"$CLAUDE_TOOL_FILE_PATH\" == *.jsx || \"$CLAUDE_TOOL_FILE_PATH\" == *.tsx ]]; then npx eslint \"$CLAUDE_TOOL_FILE_PATH\" --fix 2>/dev/null || true; elif [[ \"$CLAUDE_TOOL_FILE_PATH\" == *.py ]]; then pylint \"$CLAUDE_TOOL_FILE_PATH\" 2>/dev/null || true; fi"
+          },
+          {
+            "type": "command",
+            "command": "if [[ \"$CLAUDE_TOOL_FILE_PATH\" == *.js || \"$CLAUDE_TOOL_FILE_PATH\" == *.ts || \"$CLAUDE_TOOL_FILE_PATH\" == *.jsx || \"$CLAUDE_TOOL_FILE_PATH\" == *.tsx || \"$CLAUDE_TOOL_FILE_PATH\" == *.json || \"$CLAUDE_TOOL_FILE_PATH\" == *.css || \"$CLAUDE_TOOL_FILE_PATH\" == *.html ]]; then npx prettier --write \"$CLAUDE_TOOL_FILE_PATH\" 2>/dev/null || true; elif [[ \"$CLAUDE_TOOL_FILE_PATH\" == *.py ]]; then black \"$CLAUDE_TOOL_FILE_PATH\" 2>/dev/null || true; elif [[ \"$CLAUDE_TOOL_FILE_PATH\" == *.go ]]; then gofmt -w \"$CLAUDE_TOOL_FILE_PATH\" 2>/dev/null || true; fi"
+          }
+        ]
+      }
+    ]
+  }
+}
+```
+
+**Supported languages:**
+- JavaScript/TypeScript: ESLint + Prettier
+- Python: Pylint + Black
+- Go: gofmt
+- Rust: rustfmt
+- PHP: php-cs-fixer
+
+### 4. Completion Notifications (optional)
+
+Get notified when Claude Code finishes a task:
+
+```json
+{
+  "hooks": {
+    "Stop": [
+      {
+        "matcher": "*",
+        "hooks": [
+          {
+            "type": "command",
+            "command": "if command -v osascript >/dev/null 2>&1; then osascript -e 'display notification \"Tool: Operation completed\" with title \"Claude Code\"'; elif command -v notify-send >/dev/null 2>&1; then notify-send 'Claude Code' \"Tool: $CLAUDE_TOOL_NAME completed\"; fi"
+          },
+          {
+            "type": "command",
+            "command": "afplay /System/Library/Sounds/Glass.aiff"
+          }
+        ]
+      }
+    ]
+  }
+}
+```
+
+**Platform support:**
+- macOS: Native notifications via `osascript` + sound via `afplay`
+- Linux: Desktop notifications via `notify-send`
+
+## Components
+
+### Auto-Approve Hook
+
+PreToolUse hook that intelligently auto-approves safe commands:
+
+**Allowed by default:**
+- Read-only commands (`ls`, `cat`, `head`, `tail`, `tree`)
+- Git read operations (`git status`, `git diff`, `git log`, `git branch`)
+- Development commands (`npm test`, `npm run build`, `pnpm test`, `pytest`)
+- Search tools (`grep`, `rg`, `find`, `fd`, `jq`)
+- File system operations (`mkdir`, `touch`, `cp`, `mv`)
+- Git write operations (`git add`, `git commit`, `git push`)
+- GitHub CLI (`gh pr`, `gh issue`, `gh repo`)
+
+**Denied automatically:**
+- Dangerous commands (`sudo`, `rm -rf`, `mkfs`, `shutdown`)
+- System modifications (`chmod 777`, `chown :* /`)
+- Pipe to shell (`curl | bash`, `wget | sh`)
+- Fork bombs and kill commands
+
+**Prompted (normal permission system):**
+- Everything else defers to Claude Code's built-in permissions
+
+#### Customizing Rules
+
+Edit `.claude/scripts/auto_approve_safe.rules.json`:
+
+```json
+{
+  "allow_patterns": [
+    "^your-custom-safe-command"
+  ],
+  "deny_patterns": [
+    "^your-dangerous-command"
+  ],
+  "sensitive_paths": [
+    "\\.secret$"
+  ]
+}
+```
+
+### Context Monitor
+
+Status line script showing:
+- Current model name (e.g., `[Claude Opus 4]`)
+- Working directory
+- Git branch with visual indicator
+- Context usage bar with percentage
+- Color-coded warnings:
+  - 🟢 Green: < 50% usage
+  - 🟡 Yellow: 50-75% usage
+  - 🟠 Light red: 75-90% usage
+  - 🔴 Red: 90-95% usage
+  - 🔴 Blinking: > 95% usage (CRITICAL)
+
+### Optimize Command
+
+Run `/optimize-auto-approve-hook` to:
+1. Analyze the decision log (`.claude/auto_approve_safe.decisions.jsonl`)
+2. Validate existing ALLOW rules aren't too permissive
+3. Identify frequently-asked commands that could be safely auto-allowed
+4. Check for overly broad DENY patterns causing false positives
+5. Generate new regex patterns for safe commands
+6. Clean up and archive the decision log
+
+## Decision Logging
+
+When `ENABLE_DECISION_LOG = True` (default), the hook logs all decisions to:
+```
+.claude/auto_approve_safe.decisions.jsonl
+```
+
+Each entry includes:
+- Timestamp
+- Tool name
+- Decision (allow/deny/ask)
+- Reason for decision
+- Input summary
+
+Use `/optimize-auto-approve-hook` to analyze this log and improve your rules.
+
+## File Structure
+
+After installation, files are placed in:
+
+```
+.claude/
+├── scripts/
+│   ├── auto_approve_safe.py
+│   ├── auto_approve_safe.rules.json
+│   └── context-monitor.py
+└── commands/
+    └── optimize-auto-approve-hook.md
+```
+
+## Uninstallation
+
+```bash
+npm uninstall @torka/claude-qol
+```
+
+**Manual cleanup** (if files remain after uninstall):
+
+```bash
+rm -f .claude/scripts/auto_approve_safe.py \
+      .claude/scripts/auto_approve_safe.rules.json \
+      .claude/scripts/context-monitor.py \
+      .claude/commands/optimize-auto-approve-hook.md
+```
+
+**Note**: Your `settings.local.json` is not modified—you may want to manually remove hook/statusLine configurations.
+
+## Contributing
+
+1. Fork the repository
+2. Create a feature branch
+3. Make your changes
+4. Submit a pull request
+
+## License
+
+MIT License - see [LICENSE](LICENSE) for details.
+
+## Author
+
+Varun Torka
+
+## Links
+
+- [Claude Code Documentation](https://code.claude.com/docs)
+- [npm Package](https://www.npmjs.com/package/@torka/claude-qol)

package/commands/optimize-auto-approve-hook.md

@@ -0,0 +1,406 @@
+---
+description: 'Analyze auto-approve decisions to identify safe patterns for auto-allowing and validate existing rules'
+---
+
+# Auto-Approve Hook Optimizer
+
+You are a security-conscious CLI analyst. Analyze the auto-approve decision log to identify patterns that can be safely auto-allowed, validate existing rules, and maintain the log file.
+
+## Overview
+
+This command performs a comprehensive analysis of the auto-approve hook's decision log to:
+1. Validate that existing ALLOW rules aren't too permissive
+2. Identify frequently asked commands that could be safely auto-allowed
+3. Check for overly broad DENY patterns causing false positives
+4. Clean up the decision log after analysis
+
+## Files Involved
+
+| File | Purpose |
+|------|---------|
+| `.claude/auto_approve_safe.decisions.jsonl` | Decision log (read, then archive/clear) |
+| `.claude/scripts/auto_approve_safe.rules.json` | Rules config (read, then edit) |
+| `.claude/auto_approve_safe.decisions.archived.jsonl` | Archive file (create/append) |
+
+---
+
+## Workflow Phases
+
+<steps>
+
+### Phase 0: Pre-flight Checks
+
+1. **Read the decision log** at `.claude/auto_approve_safe.decisions.jsonl`
+   - If missing or empty: Stop with message "No decision log found. Run some operations first to generate decisions."
+
+2. **Read the rules file** at `.claude/scripts/auto_approve_safe.rules.json`
+   - If missing: Warn but continue (will create recommendations only)
+
+3. **Display summary to user:**
+   ```
+   Decision Log Summary
+   ====================
+   Total entries: {count}
+   Date range: {earliest} to {latest}
+
+   By decision type:
+   - ALLOW: {count} ({percentage}%)
+   - DENY:  {count} ({percentage}%)
+   - ASK:   {count} ({percentage}%)
+
+   By tool:
+   - Bash:  {count}
+   - Read:  {count}
+   - Write: {count}
+   - Edit:  {count}
+   - Other: {count}
+   ```
+
+### Phase 1: Parse and Categorize
+
+Parse all JSONL entries and group by decision type:
+
+**For Bash commands:**
+- Extract the `command` field from `input`
+- Normalize: collapse whitespace, extract base command and flags
+- Group similar commands (e.g., all `npm run *` together)
+
+**For File operations (Read/Write/Edit):**
+- Extract the `file_path` field from `input`
+- Identify path patterns (e.g., `src/**/*.ts`, `docs/*.md`)
+- Group by directory structure
+
+**Create internal data structures:**
+```
+allow_entries: [{ts, tool, input, reason}]
+deny_entries:  [{ts, tool, input, reason}]
+ask_entries:   [{ts, tool, input, reason}]
+```
+
+### Phase 2: Analyze ALLOW Decisions (Validate Safety)
+
+Review all ALLOW decisions and flag potentially unsafe patterns:
+
+| Red Flag | Detection Logic | Severity |
+|----------|-----------------|----------|
+| Outside project dir | Path contains `..` or absolute path not under cwd | HIGH |
+| Arbitrary code exec | `eval`, `exec`, backticks without known safe args | HIGH |
+| Network with dynamic URL | `curl`/`wget` with variable/constructed URLs | MEDIUM |
+| Dangerous flags | `--force`, `-f`, `--no-verify`, `--hard` | MEDIUM |
+| Recursive delete | `rm -r` or `rm -rf` (should be deny) | HIGH |
+| Pipe to shell | `| bash`, `| sh`, `| zsh` | HIGH |
+
+**Output format:**
+```
+POTENTIAL UNSAFE ALLOWS
+=======================
+[HIGH] Command: rm -rf node_modules
+       Reason: Matched "Matches safe allowlist" but contains recursive delete
+       Recommendation: Add to deny_patterns or remove from allow_patterns
+
+[MEDIUM] Command: curl https://example.com/script.sh
+         Reason: Network command allowed but URL could vary
+         Recommendation: Review if URL should be restricted
+```
+
+### Phase 3: Analyze DENY Decisions (Check for False Positives)
+
+Identify overly broad deny patterns that might block safe operations:
+
+**Look for:**
+- Non-recursive `rm` blocked by recursive pattern (e.g., `rm file.txt` blocked by `rm.*-rf`)
+- Safe paths incorrectly matching `sensitive_paths` (e.g., `src/utils/tokenizer.ts` matching `token`)
+- Common dev commands that should be allowed (e.g., `npm run build` blocked)
+
+**Output format:**
+```
+POTENTIAL FALSE POSITIVES
+=========================
+[ ] Pattern: "\\brm\\s+.*(-r|-rf)" may be too broad
+    Blocked: rm file.txt (no -r flag)
+    Suggestion: Use "\\brm\\s+.*\\s+(-r|-rf|--recursive)\\b" instead
+
+[ ] Pattern: "\\btoken" in sensitive_paths
+    Blocked: src/utils/tokenizer.ts
+    Suggestion: Use "\\btoken(?!izer)" or "\\btokens?\\b"
+```
+
+### Phase 4: Analyze ASK Decisions (Auto-Allow Candidates)
+
+**This is the most valuable optimization phase.**
+
+Identify commands that were asked repeatedly and could be safely auto-allowed.
+
+**Criteria for safe auto-allow:**
+
+| Criterion | Threshold | Rationale |
+|-----------|-----------|-----------|
+| Frequency | 3+ occurrences | Indicates common workflow |
+| Containment | All paths within project | No system-wide impact |
+| Predictable scope | No shell expansion, no `*` in dangerous positions | Bounded behavior |
+| Reversibility | File ops can be undone, commands are read-only or local | Low risk |
+| No dangerous flags | No `--force`, `-f` unless safe context | Explicit caution |
+
+**Process each unique ASK pattern:**
+1. Count occurrences
+2. Check all criteria
+3. If all pass, generate a specific regex pattern
+4. Include sample commands that would match
+
+**Pattern Generation Heuristics:**
+
+| Observed Pattern | Generated Regex | Notes |
+|-----------------|-----------------|-------|
+| `npx tsx script.ts` | `^npx\\s+tsx\\s+[^\\|;&]+$` | No pipes or chains |
+| `npm run custom-script` | `^npm\\s+run\\s+[\\w-]+$` | Word chars and hyphen only |
+| `cat package.json` | Already covered by existing rules | Skip |
+| `rm file.txt` (non-recursive) | `^rm\\s+(?!.*(-r\\|-rf))\\S+$` | Negative lookahead for -r |
+
+**Output format:**
+```
+AUTO-ALLOW CANDIDATES
+=====================
+[ ] 1. Pattern: "^npx\\s+tsx\\s+[^|;&]+$"
+       Matches: npx tsx script.ts, npx tsx src/test.ts
+       Occurrences: 4
+       Safety: All within project, no shell operators
+
+[ ] 2. Pattern: "^python3?\\s+[\\w/.-]+\\.py$"
+       Matches: python script.py, python3 src/tools/gen.py
+       Occurrences: 3
+       Safety: Simple script execution, no args could be dangerous
+```
+
+### Phase 5: Present Consolidated Recommendations
+
+Display a structured summary of all recommendations:
+
+```
+==============================================
+AUTO-APPROVE HOOK OPTIMIZATION RECOMMENDATIONS
+==============================================
+
+ALLOW PATTERNS TO ADD:
+[ ] 1. "^npx\\s+tsx\\s+[^|;&]+$"
+       Sample: npx tsx script.ts (4 occurrences)
+
+[ ] 2. "^python3?\\s+[\\w/.-]+\\.py$"
+       Sample: python script.py (3 occurrences)
+
+DENY PATTERNS TO FIX:
+[ ] 1. Current: "\\brm\\s+.*(-r|-rf)"
+       Replace: "\\brm\\s+.*\\s+(-r|-rf|--recursive)\\b"
+       Reason: Current pattern matches non-recursive rm
+
+SENSITIVE PATHS TO ADD:
+[ ] None identified
+
+WARNINGS (Manual Review Needed):
+! 2 potentially unsafe ALLOW decisions detected
+  See Phase 2 output for details
+```
+
+### Phase 6: User Confirmation
+
+Use `AskUserQuestion` tool with `multiSelect: true` to let user select changes:
+
+**Question 1: Pattern Selection**
+```
+header: "Patterns"
+question: "Which patterns would you like to add to the allow list?"
+options:
+  - label: "All recommended patterns"
+    description: "Add all {N} patterns identified as safe"
+  - label: "Select individually"
+    description: "Review and select patterns one by one"
+  - label: "None"
+    description: "Skip pattern changes"
+```
+
+If "Select individually" chosen, ask about each pattern.
+
+**Question 2: Log Cleanup**
+```
+header: "Log cleanup"
+question: "How should the decision log be handled after analysis?"
+options:
+  - label: "Archive"
+    description: "Move entries to .archived.jsonl file, clear active log"
+  - label: "Delete"
+    description: "Remove log file entirely (hook will recreate)"
+  - label: "Keep"
+    description: "Leave log unchanged for future analysis"
+```
+
+### Phase 7: Apply Changes
+
+Based on user selections:
+
+1. **Read current rules JSON** (if it exists)
+
+2. **For each approved pattern to add:**
+   - Append to the appropriate array (`allow_patterns`, `deny_patterns`, or `sensitive_paths`)
+   - Validate the regex is syntactically correct before adding
+
+3. **For each pattern to fix:**
+   - Find and replace the old pattern with the new one
+   - Verify the replacement was made
+
+4. **Write the updated JSON file**
+   - Use proper JSON formatting (2-space indent)
+   - Preserve comments if any (though JSON doesn't support them)
+
+5. **Validate the written file:**
+   - Read it back
+   - Parse as JSON to ensure validity
+   - If invalid, restore from backup and report error
+
+### Phase 8: Decision Log Cleanup
+
+Based on user's cleanup selection:
+
+**Archive:**
+```bash
+# Append current entries to archive
+cat .claude/auto_approve_safe.decisions.jsonl >> .claude/auto_approve_safe.decisions.archived.jsonl
+# Clear active log
+echo "" > .claude/auto_approve_safe.decisions.jsonl
+```
+
+**Delete:**
+```bash
+rm .claude/auto_approve_safe.decisions.jsonl
+# Hook will recreate on next decision
+```
+
+**Keep:**
+- No action taken
+- Log remains for future analysis
+
+### Phase 9: Summary Report
+
+Display final summary:
+
+```
+=================================
+AUTO-APPROVE OPTIMIZATION COMPLETE
+=================================
+
+Changes Applied:
+- Added {N} new allow patterns
+- Fixed {N} deny patterns
+- Added {N} sensitive paths
+
+Log Cleanup:
+- Action: {Archive|Delete|Keep}
+- Entries processed: {count}
+- Archive location: .claude/auto_approve_safe.decisions.archived.jsonl
+
+To Revert Changes:
+  git checkout .claude/scripts/auto_approve_safe.rules.json
+
+Next Steps:
+- Test the new patterns by running common commands
+- Re-run /optimize-auto-approve-hook after more usage data
+```
+
+</steps>
+
+---
+
+## Safety Rules
+
+**CRITICAL - These rules must NEVER be violated:**
+
+1. **NEVER auto-apply changes** - Always require explicit user confirmation via AskUserQuestion
+2. **NEVER remove deny patterns** without explicit warning that this could allow dangerous commands
+3. **NEVER add patterns that match outside the project directory**
+4. **Validate all regex patterns** before writing to ensure they're syntactically correct
+5. **Prefer specific patterns** over broad ones (e.g., `^npm run build$` over `^npm run .*$`)
+6. **Always provide revert instructions** so user can undo changes easily
+7. **Back up before modifying** - Read the file content before editing so it can be restored if needed
+
+---
+
+## Regex Pattern Guidelines
+
+When generating regex patterns for `allow_patterns`:
+
+| Goal | Pattern Technique | Example |
+|------|-------------------|---------|
+| Match exact command | Anchors `^...$` | `^npm run build$` |
+| Allow arguments | Character class `[^|;&]+` | `^npx tsx [^|;&]+$` |
+| Word boundary | `\\b` | `\\bgit\\b` prevents matching `digit` |
+| Exclude dangerous | Negative lookahead `(?!...)` | `^rm (?!.*-rf)` |
+| Optional spaces | `\\s+` or `\\s*` | `^npm\\s+run` |
+| Filename only | `[\\w.-]+` | `^cat [\\w.-]+$` |
+| Path characters | `[\\w/.@-]+` | `^cat [\\w/.@-]+$` |
+
+**Escape requirements in JSON:**
+- Single backslash in regex → double backslash in JSON string
+- `\s` → `\\s`
+- `\b` → `\\b`
+
+---
+
+## Error Handling
+
+| Scenario | Action |
+|----------|--------|
+| Decision log missing | Stop with helpful message |
+| Rules file missing | Continue with recommendations only |
+| Invalid JSON in rules | Report error, do not modify |
+| Regex syntax error | Skip that pattern, warn user |
+| Write permission denied | Report error with chmod suggestion |
+| Archive file locked | Report error, suggest manual cleanup |
+
+---
+
+## Example Session
+
+```
+User: /optimize-auto-approve-hook
+
+Claude: Analyzing auto-approve decision log...
+
+Decision Log Summary
+====================
+Total entries: 153
+Date range: 2026-01-15 to 2026-01-16
+
+By decision type:
+- ALLOW: 142 (92.8%)
+- DENY:  3 (2.0%)
+- ASK:   8 (5.2%)
+
+[Phase 2-4 analysis output...]
+
+AUTO-APPROVE HOOK OPTIMIZATION RECOMMENDATIONS
+==============================================
+
+ALLOW PATTERNS TO ADD:
+[1] "^npx\\s+tsx\\s+[^|;&]+$"
+    Sample: npx tsx _bmad-output/tmp/generate-theme.ts (4 occurrences)
+
+[Question] Which patterns would you like to add?
+> All recommended patterns
+
+[Question] How should the decision log be handled?
+> Archive
+
+Applying changes...
+
+AUTO-APPROVE OPTIMIZATION COMPLETE
+==================================
+
+Changes Applied:
+- Added 1 new allow pattern
+
+Log Cleanup:
+- Action: Archive
+- Entries processed: 153
+- Archive location: .claude/auto_approve_safe.decisions.archived.jsonl
+
+To Revert Changes:
+  git checkout .claude/scripts/auto_approve_safe.rules.json
+```