claude-all-hands 1.0.0 → 1.0.2

package/.claude/skills/hook-development/examples/prompt-hooks.md

@@ -1,114 +0,0 @@
-# Prompt-Based Hook Examples
-
-Prompt hooks use LLM evaluation for context-aware decisions. Recommended for complex logic.
-
-## Stop Hook: Task Completion Check
-
-```json
-{
-  "hooks": {
-    "Stop": [
-      {
-        "hooks": [
-          {
-            "type": "prompt",
-            "prompt": "Evaluate if Claude should stop working. Context: $ARGUMENTS\n\nCheck:\n1. All user-requested tasks complete\n2. No errors need addressing\n3. No obvious follow-up needed\n\nRespond: {\"decision\": \"approve\" or \"block\", \"reason\": \"explanation\"}",
-            "timeout": 30
-          }
-        ]
-      }
-    ]
-  }
-}
-```
-
-## SubagentStop: Subagent Validation
-
-```json
-{
-  "hooks": {
-    "SubagentStop": [
-      {
-        "hooks": [
-          {
-            "type": "prompt",
-            "prompt": "Evaluate if this subagent completed its task. Input: $ARGUMENTS\n\nCheck:\n- Assigned task finished\n- No errors occurred\n- Results ready for parent\n\nReturn: {\"decision\": \"approve\" or \"block\", \"reason\": \"explanation\"}"
-          }
-        ]
-      }
-    ]
-  }
-}
-```
-
-## PreToolUse: File Write Safety
-
-```json
-{
-  "hooks": {
-    "PreToolUse": [
-      {
-        "matcher": "Write|Edit",
-        "hooks": [
-          {
-            "type": "prompt",
-            "prompt": "Validate file write safety for: $ARGUMENTS\n\nCheck:\n- No system paths (/etc, /usr)\n- No credentials (.env, secrets)\n- No path traversal (..)\n- Content appropriate\n\nReturn: {\"decision\": \"approve\" or \"block\", \"reason\": \"explanation\"}"
-          }
-        ]
-      }
-    ]
-  }
-}
-```
-
-## UserPromptSubmit: Security Context
-
-```json
-{
-  "hooks": {
-    "UserPromptSubmit": [
-      {
-        "hooks": [
-          {
-            "type": "prompt",
-            "prompt": "Check if user prompt requires security guidance: $ARGUMENTS\n\nIf discussing auth, permissions, API keys, or security:\nReturn: {\"systemMessage\": \"relevant security warnings\"}\n\nOtherwise return: {}"
-          }
-        ]
-      }
-    ]
-  }
-}
-```
-
-## Response Schema
-
-Prompt hooks must return JSON:
-
-```json
-{
-  "decision": "approve|block",
-  "reason": "Explanation shown to Claude when blocked",
-  "continue": false,
-  "stopReason": "Message shown when continue is false",
-  "systemMessage": "Warning shown to user"
-}
-```
-
-**Fields**:
-- `decision`: `approve` allows, `block` prevents
-- `reason`: Shown to Claude when decision is block
-- `continue`: If false, stops Claude entirely
-- `stopReason`: Message when continue is false
-- `systemMessage`: Additional user-visible message
-
-## Prompt Variable
-
-Use `$ARGUMENTS` as placeholder for hook input JSON. If omitted, input is appended to prompt.
-
-## Best Practices
-
-1. **Be specific** - Clearly state evaluation criteria
-2. **List checks** - Numbered criteria are clearer
-3. **Require JSON response** - Explicit format in prompt
-4. **Set timeouts** - Default 30s, adjust as needed
-5. **Use for complex decisions** - Simple checks use command hooks

package/.claude/skills/hook-development/references/event-reference.md

@@ -1,226 +0,0 @@
-# Hook Event Reference
-
-Complete input/output specifications for all hook events.
-
-## Common Input Fields (All Events)
-
-All hooks receive JSON via stdin:
-
-```json
-{
-  "session_id": "abc123",
-  "transcript_path": "/path/to/transcript.jsonl",
-  "cwd": "/current/working/directory",
-  "permission_mode": "default|plan|acceptEdits|bypassPermissions",
-  "hook_event_name": "PreToolUse"
-}
-```
-
-## PreToolUse
-
-**When**: After Claude creates tool parameters, before execution
-
-**Matchers**: `Task`, `Bash`, `Read`, `Write`, `Edit`, `Glob`, `Grep`, `WebFetch`, `WebSearch`, `mcp__*`
-
-**Input**:
-```json
-{
-  "tool_name": "Write",
-  "tool_input": {
-    "file_path": "/path/to/file.txt",
-    "content": "file content"
-  },
-  "tool_use_id": "toolu_01ABC123..."
-}
-```
-
-**JSON Output (exit 0)**:
-```json
-{
-  "hookSpecificOutput": {
-    "hookEventName": "PreToolUse",
-    "permissionDecision": "allow|deny|ask",
-    "permissionDecisionReason": "Explanation",
-    "updatedInput": {
-      "file_path": "/modified/path.txt"
-    }
-  }
-}
-```
-
-**Exit 2**: Blocks tool call, stderr shown to Claude
-
-## PostToolUse
-
-**When**: After tool completes successfully
-
-**Matchers**: Same as PreToolUse
-
-**Input** (adds to PreToolUse):
-```json
-{
-  "tool_response": {
-    "filePath": "/path/to/file.txt",
-    "success": true
-  }
-}
-```
-
-**JSON Output (exit 0)**:
-```json
-{
-  "decision": "block",
-  "reason": "Explanation for Claude",
-  "hookSpecificOutput": {
-    "hookEventName": "PostToolUse",
-    "additionalContext": "Additional info for Claude"
-  }
-}
-```
-
-**Exit 2**: Shows stderr to Claude (tool already ran)
-
-## UserPromptSubmit
-
-**When**: User submits prompt, before Claude processes
-
-**Input**:
-```json
-{
-  "prompt": "Write a function to calculate factorial"
-}
-```
-
-**Output Options**:
-1. Plain stdout (exit 0): Added as context
-2. JSON with `additionalContext`: Structured context injection
-3. JSON with `decision: "block"`: Prevents prompt processing
-
-```json
-{
-  "decision": "block",
-  "reason": "Shown to user (not Claude)",
-  "hookSpecificOutput": {
-    "hookEventName": "UserPromptSubmit",
-    "additionalContext": "Context added to conversation"
-  }
-}
-```
-
-**Exit 2**: Blocks prompt, erases it, shows stderr to user only
-
-## Stop
-
-**When**: Main agent finishes responding (not on user interrupt)
-
-**Input**:
-```json
-{
-  "stop_hook_active": true
-}
-```
-
-Note: `stop_hook_active` is true when Claude is continuing from a previous stop hook. Check this to prevent infinite loops.
-
-**JSON Output (exit 0)**:
-```json
-{
-  "decision": "block",
-  "reason": "Must continue because: [explanation for Claude]"
-}
-```
-
-**Exit 2**: Blocks stopping, stderr shown to Claude
-
-## SubagentStop
-
-**When**: Subagent (Task tool) finishes
-
-**Input**: Same as Stop
-
-**Output**: Same as Stop
-
-## SessionStart
-
-**When**: New session or resume
-
-**Matchers**: `startup`, `resume`, `clear`, `compact`
-
-**Input**:
-```json
-{
-  "source": "startup|resume|clear|compact"
-}
-```
-
-**Special**: Write to `$CLAUDE_ENV_FILE` to persist env vars:
-```bash
-echo 'export MY_VAR=value' >> "$CLAUDE_ENV_FILE"
-```
-
-**JSON Output (exit 0)**:
-```json
-{
-  "hookSpecificOutput": {
-    "hookEventName": "SessionStart",
-    "additionalContext": "Context loaded at startup"
-  }
-}
-```
-
-Plain stdout also added as context.
-
-## SessionEnd
-
-**When**: Session ends
-
-**Input**:
-```json
-{
-  "reason": "clear|logout|prompt_input_exit|other"
-}
-```
-
-Cannot block termination. Use for cleanup/logging.
-
-## PreCompact
-
-**When**: Before context compaction
-
-**Matchers**: `manual` (from `/compact`), `auto` (context full)
-
-**Input**:
-```json
-{
-  "trigger": "manual|auto",
-  "custom_instructions": ""
-}
-```
-
-## Notification
-
-**When**: Claude sends notifications
-
-**Matchers**: `permission_prompt`, `idle_prompt`, `auth_success`, `elicitation_dialog`
-
-**Input**:
-```json
-{
-  "message": "Claude needs your permission to use Bash",
-  "notification_type": "permission_prompt"
-}
-```
-
-Output logged to debug only.
-
-## Matcher Patterns
-
-```json
-"matcher": "Write"           // Exact match
-"matcher": "Write|Edit"      // Multiple tools
-"matcher": "*"               // All tools (or omit)
-"matcher": "mcp__memory__.*" // Regex: all memory MCP tools
-"matcher": "mcp__.*__delete" // Regex: all MCP delete tools
-```
-
-Case-sensitive. Supports regex.

package/.claude/skills/repomix-extraction/SKILL.md

@@ -1,89 +0,0 @@
----
-name: repomix-extraction
-description: Directory extraction using repomix for pattern discovery. Use when skills/directives fall short and you need to understand codebase patterns in a specific directory.
----
-
-# Repomix Extraction
-
-Pack directory contents into AI-friendly format for comprehensive pattern analysis.
-
-## When to Use
-
-- Skills/directives don't cover the directory you need to understand
-- Need to discover patterns in unfamiliar codebase areas
-- Creating specialist agents that need repo pattern knowledge
-- Analyzing code structure before making changes
-
-## Quick Reference
-
-```bash
-# Pack directory (outputs repomix-output.xml)
-npx repomix@latest path/to/directory
-
-# Include specific patterns
-npx repomix@latest --include "**/*.ts,**/*.md" path/to/directory
-
-# Exclude noise
-npx repomix@latest --ignore "**/*.log,node_modules/" path/to/directory
-
-# Compress for large directories (tree-sitter extraction)
-npx repomix@latest --compress path/to/directory
-
-# Output to stdout (for piping)
-npx repomix@latest --stdout path/to/directory
-```
-
-## Extraction Process
-
-### 1. Pack Directory
-```bash
-npx repomix@latest path/to/directory
-```
-Output: `repomix-output.xml` in current directory
-
-### 2. Read Output
-Read the XML file to analyze:
-- File structure conventions
-- Directory organization patterns
-- Naming conventions (files, functions, variables)
-- Code style/idioms
-- Import/export patterns
-- Error handling approaches
-
-### 3. Synthesize Patterns
-Extract actionable patterns:
-- Recurring code patterns
-- Domain-specific conventions
-- Anti-patterns to avoid
-- Guidelines for system prompts or implementation
-
-## Options Reference
-
-| Flag | Use |
-|------|-----|
-| `--include "glob"` | Include specific file patterns |
-| `--ignore "glob"` | Exclude patterns |
-| `--compress` | Tree-sitter extraction (large dirs) |
-| `--stdout` | Output to stdout instead of file |
-| `--remote URL` | Pack remote repo (GitHub shorthand) |
-
-## Output Format
-
-Return extracted patterns as:
-```markdown
-## Directory Analysis: [path]
-
-### Structure
-- [directory organization patterns]
-
-### Conventions
-- Naming: [patterns]
-- Style: [patterns]
-
-### Key Patterns
-- [pattern 1]
-- [pattern 2]
-
-### Recommendations
-[Actionable guidance based on patterns]
-```

package/.claude/skills/skill-development/SKILL.md

@@ -1,236 +0,0 @@
----
-name: skill-development
-description: Use when user asks to "create a skill", "build a new skill", "write skill instructions", "improve skill description", "organize skill content", or needs guidance on skill structure, progressive disclosure, or skill development best practices. (project)
----
-
-# Skill Development
-
-Guide for creating effective Claude Code skills with proper structure, progressive disclosure, and validation.
-
-## Overview
-
-Skills extend Claude's capabilities through specialized knowledge and workflows. Each skill packages expertise into a discoverable capability that Claude invokes autonomously based on description triggers.
-
-### Skill Anatomy
-
-```
-skill-name/
-├── SKILL.md (required)
-│   ├── YAML frontmatter (name, description)
-│   └── Markdown body (~500-700 words target)
-└── Bundled Resources (only if needed)
-    ├── references/     - Detailed docs, loaded as needed
-    ├── examples/       - Working code, copyable
-    └── scripts/        - Utilities, executable
-```
-
-### Progressive Disclosure
-
-Three-level loading for context efficiency:
-
-| Level | Contents | When Loaded | Size Target |
-|-------|----------|-------------|-------------|
-| Metadata | name + description | Always | ~100 words |
-| Body | SKILL.md content | Skill triggers | ~500-700 words |
-| Resources | references/, examples/, scripts/ | As needed | Unlimited |
-
-For detailed patterns, see `references/progressive-disclosure.md`.
-
-## Creation Process
-
-### Step 1: Understand Use Cases
-
-Before writing, clarify concrete usage scenarios:
-
-- What specific phrases trigger this skill?
-- What tasks does it help accomplish?
-- What would a user say to invoke it?
-
-### Step 2: Plan Reusable Contents
-
-Analyze each use case to identify:
-
-1. **Scripts**: Code rewritten repeatedly or needing deterministic reliability
-2. **References**: Documentation Claude should consult while working
-3. **Assets**: Files used in output (templates, images)
-
-### Step 3: Create Structure
-
-```bash
-mkdir -p .claude/skills/skill-name
-touch .claude/skills/skill-name/SKILL.md
-```
-
-**Minimal by default**: Only create `references/`, `examples/`, or `scripts/` when content exceeds SKILL.md capacity or requires separate files. Most skills need only SKILL.md.
-
-### Step 4: Write SKILL.md
-
-#### Frontmatter Requirements
-
-```yaml
----
-name: skill-name          # lowercase-hyphenated, max 64 chars
-description: [trigger description, max 1024 chars]
----
-```
-
-**Description format** (direct with triggers):
-```yaml
-description: Use when user asks to "create X", "configure Y", "build Z", or needs guidance on [domain]. Provides [capability summary].
-```
-
-#### Body Writing Style
-
-Use **imperative/infinitive form** throughout:
-
-| Correct | Incorrect |
-|---------|-----------|
-| Create the configuration file | You should create the configuration file |
-| Validate input before processing | You need to validate input |
-| Parse frontmatter using sed | Claude should parse frontmatter |
-
-For complete style rules, see `references/writing-style.md`.
-
-#### Body Structure
-
-```markdown
-# Skill Name
-
-Overview (2-3 sentences).
-
-## Quick Reference
-Essential patterns, commands, key concepts.
-
-## Key Workflows
-Step-by-step procedures for main use cases.
-
-## Additional Resources
-
-### Reference Files
-- **`references/detailed-guide.md`** - Extended documentation
-
-### Examples
-- **`examples/working-example.sh`** - Annotated example
-```
-
-**Keep body lean**: Only move to references/ when content exceeds ~700 words or requires separate file format.
-
-### Step 5: Validate
-
-**Structure**:
-- [ ] SKILL.md exists with valid YAML frontmatter
-- [ ] `name` field: lowercase-hyphenated, max 64 chars
-- [ ] `description` field: max 1024 chars
-- [ ] Referenced files exist
-
-**Description Quality**:
-- [ ] Direct format ("Use when...")
-- [ ] Specific trigger phrases ("create X", "configure Y")
-- [ ] Not vague or generic
-
-**Content Quality**:
-- [ ] Body uses imperative/infinitive form (not second person)
-- [ ] Body is ~500-700 words (move excess to references/)
-- [ ] Resources only created when necessary
-- [ ] Resources clearly referenced in body if present
-
-### Step 6: Iterate
-
-After using the skill on real tasks:
-
-1. Notice struggles or inefficiencies
-2. Identify improvements to SKILL.md or resources
-3. Implement changes
-4. Test again
-
-## Degrees of Freedom
-
-Match constraint level to task criticality:
-
-| Level | When to Use | Example |
-|-------|-------------|---------|
-| High (text) | Multiple valid approaches | "Generate appropriate tests" |
-| Medium (parameterized) | Preferred pattern exists | Script with configurable params |
-| Low (exact) | Fragile/critical ops | Exact shell commands |
-
-## Common Mistakes
-
-### Weak Trigger Description
-
-**Bad**:
-```yaml
-description: Provides guidance for working with hooks.
-```
-
-**Good**:
-```yaml
-description: Use when user asks to "create a hook", "add a PreToolUse hook", "validate tool use". Provides hooks API guidance.
-```
-
-### Too Much in SKILL.md
-
-**Bad**: 1,000+ words in SKILL.md without references
-
-**Good**: ~500-700 words in SKILL.md, move excess to references/
-
-### Unnecessary Reference Files
-
-**Bad**: Creating references/ for simple skills that fit in ~500 words
-
-**Good**: Most skills are SKILL.md only. Create subdirs only when:
-- Content exceeds ~700 words
-- Need executable scripts or separate file formats
-- Multiple distinct reference documents required
-
-### Second Person Writing
-
-**Bad**: "You should start by reading the configuration."
-
-**Good**: "Start by reading the configuration."
-
-## Quick Reference Structures
-
-### Minimal Skill
-```
-skill-name/
-└── SKILL.md
-```
-
-### Standard Skill (Recommended)
-```
-skill-name/
-├── SKILL.md
-├── references/
-│   └── detailed-guide.md
-└── examples/
-    └── working-example.sh
-```
-
-### Complete Skill
-```
-skill-name/
-├── SKILL.md
-├── references/
-│   ├── patterns.md
-│   └── advanced.md
-├── examples/
-│   ├── example1.sh
-│   └── example2.json
-└── scripts/
-    └── validate.sh
-```
-
-## Additional Resources
-
-### Reference Files
-
-- **`references/progressive-disclosure.md`** - Directory structure patterns, when to use each level
-- **`references/writing-style.md`** - Imperative form rules, formatting best practices
-
-### Examples
-
-- **`examples/complete-skill-examples.md`** - Annotated complete skills with commentary
-
-### Scripts
-
-- **`scripts/validate-skill.sh`** - Validate skill file structure and content