rlm

Process large codebases with parallel agents using map-reduce.

145|28|Updated Jan 31, 2026
One-click install
npx skills add https://github.com/guia-matthieu/clawfu-skills --skill rlm-guia-matthieu
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: rlm
Source: https://github.com/guia-matthieu/clawfu-skills/tree/main/skills/meta/rlm
Command: npx skills add https://github.com/guia-matthieu/clawfu-skills --skill rlm-guia-matthieu

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires rlm.py, and includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill tackles the challenge of analyzing and processing massive codebases that exceed typical context window limitations, preventing context rot and enabling deep analysis.

Core Features & Use Cases

  • Recursive Analysis: Employs a map-reduce pattern with parallel agents to process code in manageable chunks.
  • Context Management: Treats context as an external resource, ensuring agents work with fresh information.
  • Use Case: Analyze a project with over 100 files to find all instances of a specific API endpoint and verify if authentication is applied to each.

Quick Start

Use the rlm skill to find all usages of the 'getUser' function across the entire project.

Frequently Asked Questions about rlm

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I analyze a large codebase that exceeds the context window limit?

To analyze a large codebase exceeding context limits, you can use the Recursive Language Model pattern to treat code as an external environment. This orchestrates parallel background agents using a map-reduce pattern to process files in manageable chunks, preventing context degradation.

What is the best way to find all usages of a specific function across a project with over 100 files?

The best way to find function usages across a massive project is applying a map-reduce pattern with parallel agents. This approach slices the large codebase programmatically, ensuring agents work with fresh context to accurately locate every instance of the target function.

Can I use bash tools like find and grep for context management on massive codebases?

Yes, you can use bash tools like find and grep for context management on massive codebases. A native mode supports these standard command-line utilities, while a strict mode utilizes Python scripts for dense data analysis and programmatic slicing of large files.

How does treating context as an external resource prevent context rot during code analysis?

Treating context as an external resource prevents context rot by ensuring parallel agents process code in isolated, manageable chunks. This map-reduce orchestration feeds agents fresh information, avoiding the degradation that occurs when exceeding typical context window limitations.

Does this map-reduce approach require Python scripts for programmatic file slicing?

No, Python scripts are not strictly required for file slicing. You can operate in native mode using standard bash utilities like find and grep, or switch to strict mode with Python scripts when you need dense data analysis and programmatic slicing of large files.