prompt-riser

CLI that runs prompts through multiple reasoning strategies and visualizes the reasoning paths as a tree

prompt-riser is a reasoning debugger CLI. It runs one prompt through multiple reasoning strategies, parses each reasoning trace, and renders the results as a terminal tree, JSON, or Mermaid flowchart.

Install

pip install -e .

Usage

prompt-riser run "What's the best way to deploy on a budget?"
prompt-riser run "Explain quantum computing" --strategies cot,self-correct
prompt-riser run "Design a REST API" --model anthropic/claude-sonnet-4 --base-url https://api.openrouter.ai/v1
prompt-riser run "Solve this equation" --output json
prompt-riser run "Debug this code" --verbose

Commands

  • prompt-riser run <prompt>: execute strategies and render results
  • prompt-riser list-strategies: list available strategy names and descriptions
  • prompt-riser config: print effective configuration
  • prompt-riser config init: create default config at ~/.config/prompt-riser/config.yaml

Strategies

Strategy Description
cot Chain-of-Thought: numbered, step-by-step reasoning
self-correct Self-correction loop: generate, critique, improve
step-back Step-back: derive broader principle, then answer specifics
constraint Constraint-based reasoning with explicit constraints
tree-of-thoughts Explore branches, evaluate, select best

Output Formats

  • tree (default): Rich terminal tree with colored nodes
  • json: Structured data for programmatic use
  • mermaid: Mermaid.js flowchart syntax

Configuration

Config at ~/.config/prompt-riser/config.yaml:

api_key: ""
base_url: "https://api.openai.com/v1"
model: "gpt-4o"
default_strategies: ["cot", "self-correct", "step-back"]
default_passes: 3
default_branches: 3

OPENAI_API_KEY is used when api_key is missing.