Just Prompt for Claude Code: Setup & Config

What you can do with Just Prompt in Claude Code

Keep model comparison, saved answers, and a panel's final decision within one MCP workflow.

• Compare independently generated answers to one prompt without switching assistant integrations.
• Discover provider-specific model choices before composing a multi-model request.

Before you start: Just Prompt in Claude Code

• Python and uv
• Local checkout with dependencies prepared using uv sync
• Credentials for chosen hosted providers, or an Ollama endpoint

This is the minimal README launch, not uv sync alone, which only prepares dependencies. The dot refers to the checkout directory; use the checkout's path if the client starts elsewhere. Provider values can be supplied by environment or .env. The optional --default-models argument chooses fallback models.

Just Prompt Claude Code configuration

1. Choose Claude Code configuration scope

   For Just Prompt, open Project scope: .mcp.json at the repository root. Project JSON: merge into mcpServers; user scope: use the CLI.

2. Add Just Prompt to Claude Code

   Use the just-prompt entry below in mcpServers. Keep Just Prompt's documented stdio connection fields and your other Claude Code entries.

3. Check Just Prompt in Claude Code

   Just Prompt: Inspect claude mcp list and /mcp status.

{
  "mcpServers": {
    "just-prompt": {
      "type": "stdio",
      "command": "uv",
      "args": [
        "--directory",
        ".",
        "run",
        "just-prompt"
      ],
      "env": {
        "OPENAI_API_KEY": "<credential if selecting OpenAI models>",
        "ANTHROPIC_API_KEY": "<credential if selecting Anthropic models>",
        "GEMINI_API_KEY": "<credential if selecting Gemini models>",
        "GROQ_API_KEY": "<credential if selecting Groq models>",
        "DEEPSEEK_API_KEY": "<credential if selecting DeepSeek models>",
        "OLLAMA_HOST": "<address of the Ollama instance to consult>"
      }
    }
  }
}

claude mcp add-json --scope project 'just-prompt' '{"type":"stdio","command":"uv","args":["--directory",".","run","just-prompt"],"env":{"OPENAI_API_KEY":"<credential if selecting OpenAI models>","ANTHROPIC_API_KEY":"<credential if selecting Anthropic models>","GEMINI_API_KEY":"<credential if selecting Gemini models>","GROQ_API_KEY":"<credential if selecting Groq models>","DEEPSEEK_API_KEY":"<credential if selecting DeepSeek models>","OLLAMA_HOST":"<address of the Ollama instance to consult>"}}'

Just Prompt tools available to Claude Code

Just Prompt's stored tool list documents 6 tools; compare it with Claude Code's connected tool list, because the stored list is not a live check.

ceo_and_board

For Just Prompt's ceo_and_board, review the permission request and arguments in Claude Code before approval.

Gather a panel's model responses, ask a designated model for a decision, and save the resulting files.

prompt_from_file_to_file

For Just Prompt's prompt_from_file_to_file, review the permission request and arguments in Claude Code before approval.

Read a local prompt by absolute path and write model answers to Markdown output files.

prompt_from_file

For Just Prompt's prompt_from_file, review the permission request and arguments in Claude Code before approval.

prompt

For Just Prompt's prompt, review the permission request and arguments in Claude Code before approval.

list_models

For Just Prompt's list_models, review the permission request and arguments in Claude Code before approval.

Discover model choices under one specified provider.

list_providers

For Just Prompt's list_providers, review the permission request and arguments in Claude Code before approval.

Just Prompt: full stored tool descriptions

Troubleshooting Just Prompt in Claude Code

• If Just Prompt fails to start, check uv, the working directory, and Claude Code's selected scope; inspect claude mcp list and /mcp.
• Just Prompt: check ANTHROPIC_API_KEY, DEEPSEEK_API_KEY, GEMINI_API_KEY, GROQ_API_KEY, OLLAMA_HOST, OPENAI_API_KEY; keep credentials out of shared .mcp.json.
• Just Prompt uses stdio here: Claude Code must keep the configured process running; protocol output belongs on stdout and diagnostics on stderr.
• Known limitation: File-based operations require absolute prompt and output paths.
• Known limitation: Each hosted backend needs its own usable credential.
• Known limitation: Reasoning suffixes are tied to documented models and provider-specific limits.
• General Claude Code troubleshooting lives in the /clients/claude-code guide.

Frequently asked questions