rlm-pattern

rlm-pattern is a skill for Claude Code from zircote-plugins/claude-team-orchestration. It costs 55 tokens per session (7,631 once invoked), scanned A, original, MIT.

A method for analyzing files that are too large for one AI request by splitting them into smaller parts and using multiple analyst agents. Their findings are then combined into one report.

In plain words
What is it for?
Use it to chunk and analyze large files, documents, or tables, then produce a consolidated analysis.
Why use it?
It avoids context-size limits and lets large documents or datasets be examined without loading everything at once.

Skill for Claude Code

Written for Claude Code: user-invocable in frontmatter. Also seen: mentions subagents; mentions Claude Code.

Part of the swarm plugin — 10 skills, 5 agents shipped together

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/zircote-plugins/claude-team-orchestration/rlm-pattern
Any agent
npx skills add zircote-plugins/claude-team-orchestration --skill rlm-pattern
Clone the repo
git clone --depth 1 https://github.com/zircote-plugins/claude-team-orchestration

Made for: Claude Code.

Or install swarm, the plugin that ships this one along with the rest of its 10 skills, 5 agents.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for rlm-pattern

README.md
[![agentmods](https://agentmods.dev/badge/skills/zircote-plugins/claude-team-orchestration/rlm-pattern.svg)](https://agentmods.dev/skills/zircote-plugins/claude-team-orchestration/rlm-pattern)
Your own site
<a href="https://agentmods.dev/skills/zircote-plugins/claude-team-orchestration/rlm-pattern"><img src="https://agentmods.dev/badge/skills/zircote-plugins/claude-team-orchestration/rlm-pattern.svg" alt="Measured on agentmods" height="20"></a>
Per session 55 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 7,631 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00055 $0.07631
Opus 5 $0.00028 $0.03816
Sonnet 5 $0.00011 $0.01526
Haiku 4.5 $0.00006 $0.00763

Measured 6d ago against content hash ef6810bad514, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-06, from the pricing page.

Security

Grade A, and why

rlm-pattern scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 6d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

skills/rlm-pattern/SKILL.md · 569 lines

How it starts

The opening of the file, as written. The whole thing — 569 lines — stays where its author put it; the contents beside it link to each section on GitHub.

RLM (Recursive Language Model) Pattern

Process files that exceed context limits by dividing them into partitions and coordinating a team of analyst agents to analyze each partition in parallel, then synthesizing their reports.

Based on: arXiv:2512.24601


The Pattern

  1. Team Lead detects content type, determines partitioning strategy, and divides the content into manageable chunks.

  2. Each Analyst agent (selected by content type) independently analyzes their assigned partition and reports back to team-lead with structured findings.

  3. Team Lead collects all analyst reports, then either synthesizes directly or spawns a dedicated Synthesizer agent to produce the consolidated report.

This is a fan-out/fan-in coordination pattern built on agent teams.


Critical: What RLM is NOT

RLM partitions the data, not the analysis. Every analyst receives the same query and examines all columns/fields in their chunk of rows. The analysis goals are the query — not separate tasks.

WRONG (thematic decomposition):

  • Task 1: "Analyze error patterns" (reads entire file)
  • Task 2: "Analyze root causes" (reads entire file again)
  • Task 3: "Analyze trends" (reads entire file again)
  • Task 4: Synthesize (blocked by 1-3)

RIGHT (data partitioning):

  • Task 1: "Chunk 1 of 10 — rows 1-5000" (query: find error patterns, root causes, and trends)
  • Task 2: "Chunk 2 of 10 — rows 5001-10000" (same query)
  • ...
  • Task 10: "Chunk 10 of 10 — rows 45001-50000" (same query)
  • Synthesis: aggregate all chunk findings

If you find yourself creating tasks named after analysis concerns rather than data partitions, STOP — you are not doing RLM.


When to Use

  • File exceeds ~1500 lines or ~1500 CSV rows
  • You need comprehensive analysis of the full content, not just a snippet
  • Log analysis, data exports, large codebase review, document analysis
  • Grep alone can't answer the question (need contextual analysis, not keyword matching)
  • Directory contains multiple files needing cross-file analysis (see Multi-File Directory Analysis)

Read the full file on GitHub · 569 lines

Files

What ships with it

2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 6d ago First seen · 569 lines · 55 tokens per session scan A ef6810bad514

Subscribe to this mod's changes

rlm-pattern is a skill published in the GitHub repository zircote-plugins/claude-team-orchestration (14 stars, last pushed 1mo ago), licensed MIT. It adds 55 tokens to every session and 7,631 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Related

Other skills, from other repositories

content-extraction

Extracts and organizes content ideas from long-form content into detailed platform-specific tables. Use when repurposing transcripts, articles, blog posts, or any long-form content into multiple pieces across platforms.

az9713/claude-cowork-content-plugin · 43 tokens

title-generator

Generates compelling, click-worthy titles from content across multiple styles and platforms. Use when naming blog posts, videos, newsletters, social posts, or any content that needs a great headline.

az9713/claude-cowork-content-plugin · 40 tokens

linkedin-post

Generates engaging LinkedIn posts from long-form content. Use when creating LinkedIn content, building thought leadership, or repurposing content for professional audiences.

az9713/claude-cowork-content-plugin · 34 tokens

twitter-thread

Generates engaging Twitter/X threads from long-form content. Use when creating Twitter content, building audience, or breaking down complex topics into tweet-sized insights.

az9713/claude-cowork-content-plugin · 33 tokens

key-takeaways

Summarizes key takeaways from long-form content into structured bullet lists and actionable insights. Use when creating show notes, content summaries, SEO descriptions, or quick reference guides.

az9713/claude-cowork-content-plugin · 40 tokens

quote-extractor

Extracts the most quotable, shareable statements from content and formats them for social media use. Use when finding pull quotes, social proof, or highlight-worthy statements.

az9713/claude-cowork-content-plugin · 38 tokens