rlm

Decompose large inputs recursively to extend model context for code analysis.

89|5|Updated Jan 21, 2026

Search Tags:#codebase #cli #code-analysis #diff-review #recursive #rlm

Ready to run

npx skills add https://github.com/rawwerks/rlm-cli --skill rlm-rawwerks

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: rlm
Source: https://github.com/rawwerks/rlm-cli/tree/main/skills/rlm
Command: npx skills add https://github.com/rawwerks/rlm-cli --skill rlm-rawwerks

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Large language models struggle with extremely long inputs. RLM CLI solves this by decomposing inputs into smaller parts and recursively calling itself to maintain context and derive a cohesive result.

Core Features & Use Cases

Recursive decomposition of inputs to extend context beyond model limits.
Support for files, directories, URLs, and stdin to analyze codebases and diffs.
Use cases: code analysis, diff reviews, and codebase exploration at scale.

Quick Start

Run rlm ask on a large repository to analyze architecture.

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 model's context window limit?▼

Codebase analysis beyond context limits is handled by recursively decomposing inputs into smaller chunks. This maintains context across files and directories to derive cohesive architectural analysis results.

How does recursive code analysis work for reviewing large diffs?▼

Recursive diff review works by breaking large inputs into manageable parts and calling itself to process them. This extends model context to review code changes at scale across various input sources.

Can I use stdin and URLs as input sources for codebase exploration?▼

Yes, codebase exploration supports files, directories, URLs, and stdin as input sources. You can invoke commands to search and index these inputs to analyze architecture across different formats.

What is the best way to break large contexts into actionable analysis?▼

Breaking large contexts into actionable analysis is achieved through recursive decomposition. By calling itself to process smaller parts, it extends context limits and produces cohesive results for codebases.

Do I need a specific configuration file to start analyzing code with recursive commands?▼

A SKILL.md frontmatter with name and description is required to begin. The body must instruct how to invoke commands like ask, complete, search, and index, alongside optional scripts and assets directories.

When should I not use recursive decomposition for code analysis?▼

Recursive decomposition should be avoided for small codebases fitting within standard model context limits. It is specifically designed to handle extremely long inputs exceeding standard processing capabilities.