you type:   "fix this null pointer crash"

claude sees: "Work through this using the ReAct pattern — alternate between
              Thought (reasoning) and Action (what you would do), then give
              a final Answer.

              fix this null pointer crash"

git clone https://github.com/AlanRoybal/prompte-mcp
cd prompte-mcp
node bin/setup.js

node bin/setup.js --yes       # accept all defaults, no prompts
node bin/setup.js --dry-run   # preview changes without writing anything

{
  "mcpServers": {
    "prompte": {
      "command": "node",
      "args": ["/path/to/prompte-mcp/bin/prompte-mcp.js"]
    }
  },
  "hooks": {
    "UserPromptSubmit": [
      {
        "matcher": "",
        "hooks": [
          {
            "type": "command",
            "command": "python3 /path/to/prompte-mcp/hooks/user-prompt-submit.py"
          }
        ]
      }
    ]
  }
}

your prompt
    │
    ▼
┌─────────────┐
│  Classifier │  keyword heuristics (no API call)
│             │  → intent: debugging / reasoning / generation / ...
└──────┬──────┘
       │  technique affinity scores
       ▼
┌─────────────┐
│   Scorer    │  affinity × your learned acceptance rate
│             │  → selects best technique
└──────┬──────┘
       │  techniqueInstruction
       ▼
  Claude Code  ←── applies the technique using its own intelligence
       │
       ▼
    response

* just answer this exactly as asked

I've selected the Chain of Thought technique for this (debugging).

Original: why does my function crash when the list is empty?

Enhanced: Think through this step-by-step before giving your final
          answer. Show your reasoning explicitly.

          why does my function crash when the list is empty?

[A] Accept   [E] Edit   [S] Skip   [Q] Quit

{
  "enabled": true,
  "autoAccept": false,
  "bypassPrefix": "*",
  "preferredTechniques": [],
  "disabledTechniques": [],
  "llmClassifier": true,
  "maxPromptLength": 4000
}

prompte-mcp/
├── bin/
│   ├── prompte-mcp.js      MCP server
│   └── setup.js            setup script
├── src/
│   ├── classifier/         intent classification (LLM + keyword fallback)
│   ├── techniques/         8 technique definitions
│   ├── engine/             classify → score → select → rewrite
│   └── config/             ~/.prompte/ storage and acceptance rate learning
├── hooks/
│   └── user-prompt-submit.py   UserPromptSubmit hook
└── CLAUDE.md               tells Claude when to call enhance_prompt