rlm-reasoning-skill

What problem does it solve?

This Skill helps you solve complex, multi-step problems by combining iterative code execution with recursive language-model reasoning.

Core Features & Use Cases

• REPL-driven reasoning loop: Execute Python in repl blocks, inspect stdout, and refine outputs across iterations.
• Recursive decomposition: Use llm_query for simple sub-tasks and rlm_query to spawn deeper recursive RLM reasoning when code-backed iteration is needed.
• Context-aware processing: Read and transform the user input from the REPL namespace variable context using code before any final response.
• Clean completion signaling: End with exactly one FINAL(...) or FINAL_VAR(...) to reliably return the final answer.

Quick Start

Ask your AI to solve the problem by writing Python in repl blocks, using recursive sub-queries when necessary, and returning the result via FINAL or FINAL_VAR.