Chunk large text files and delegate analysis to sub-LLMs.
npx skills add https://github.com/Whamp/pi-rlm --skill rlm-whamp
Please help me install this Agent Skill. Skill: rlm Source: https://github.com/Whamp/pi-rlm/tree/main/skills/rlm Command: npx skills add https://github.com/Whamp/pi-rlm --skill rlm-whamp
The RLM skill enables processing of files that are far larger than the context window of standard language models, such as multi‑megabyte logs, extensive codebases, or whole textbooks, without exhausting token limits.
llm_query and Level 3 synthesis through subagents, keeping the main context clean.Run /skill:rlm context=path/to/large.txt query='Summarize the document' to begin processing.
High-intent search queries and answers about installing and using this skill.
To analyze large text files that exceed LLM context limits, this skill automatically splits content at logical boundaries like markdown headers or code symbols into manageable chunks, then delegates processing to sub-LLMs to keep the main context clean.
The best way to process multi-megabyte log dumps is by loading the entire document into a persistent Python REPL for structural analysis, using a handle system for token-efficient searches, and synthesizing final answers through subagents.
You can review an extensive codebase without exhausting token limits by splitting the content at logical code symbols and escalating semantic reasoning to Level 2 sub-LLMs, which keeps the main context clean while analyzing the repository.
The handle system performs token-efficient searches and incremental materialization of results, allowing you to query massive documents without loading the entire content into the context window at once.
Chunking supports splitting content at logical boundaries for markdown headers, paragraphs, JSON arrays and objects, or code symbols, enabling manageable processing for various massive document types.
legoutProcess large text and code files exceeding LLM context windows via recursive chunking and sub-agent delegation.
rafaelromaoGenerate an execution-ready plan from issue context and code exploration.
brainqub3Execute recursive language model loops for long-context tasks via a local Python REPL.
egoughnourChunk large inputs exceeding context limits and aggregate sub-query results.