v1.2.1 - RLM Examples Guide

Added

• RLM Examples Guide (docs/rlm-examples.md): Copy-paste example prompts for all 6 RLM modes with feature comparison table and sizing guidelines
  • Basic RLM (logs, text)
  • Content-Aware: Source Code (function-boundary chunks, analysis focus options)
  • Content-Aware: CSV/TSV (header-preserving, distribution-aware)
  • Content-Aware: JSON/JSONL (schema-aware, drift detection)
  • Directory Analysis (same-type files)
  • Multi-Type Directory (mixed analysts, two-phase synthesis)

Full Changelog: v1.2.0...v1.2.1

v1.2.0 - Content-Aware and Multi-File RLM

Added

• Content-Aware RLM: Auto content-type detection (extension + sniffing) and type-specific chunking
• New agents: swarm:rlm-code-analyzer, swarm:rlm-data-analyzer, swarm:rlm-json-analyzer
• Multi-File Directory RLM: Process entire directories with mixed content types in one session
  • Tiered partition budget with 30-partition global cap
  • Small file batching, two-phase synthesis, findings-in-task-descriptions pattern
• Workflow 5: Multi-File Directory Analysis end-to-end example

Changed

• Updated RLM skill with content-type detection, agent routing, and multi-file sections
• Updated all 4 analyst agents with multi-file mode workflow
• Updated synthesizer with TaskGet/TaskUpdate tools and two-phase synthesis
• Added analyst model override guardrails

v1.1.0 - RLM Pattern

Added

• RLM Pattern (Pattern 7): Fan-out/fan-in chunked parallel analysis for files exceeding context limits
• New agents: swarm:rlm-chunk-analyzer (Haiku), swarm:rlm-synthesizer (Sonnet)
• New skill: swarm:rlm-pattern — complete RLM workflow with partitioning strategies
• Workflow 4: RLM Document Analysis end-to-end example

Changed

• Updated orchestration pattern count from 6 to 7 across all documentation
• Added RLM agents to agent type selection guides

v1.0.0 — Initial Release

swarm v1.0.0

Multi-agent orchestration plugin for Claude Code. Coordinate teams of agents with shared tasks, messaging, and proven patterns.

What's included

8 modular skills under the swarm: namespace:

Skill	Description
swarm:orchestrating	Primitives overview, lifecycle, quick reference
swarm:team-management	Create teams, spawn teammates, delegate mode, shutdown
swarm:task-system	Tasks, dependencies, claiming, progress tracking
swarm:messaging	DMs, broadcasts, shutdown requests, plan approvals
swarm:agent-types	Built-in + plugin agents with selection guide
swarm:orchestration-patterns	6 patterns: parallel, pipeline, swarm, research+impl, plan approval, refactoring
swarm:spawn-backends	In-process, tmux, iTerm2 with auto-detection
swarm:error-handling	Common errors, hooks, limitations, recovery

Documentation:

• Getting Started - Install to first team run
• Agent Types - Which agent for which task
• Patterns - When to use which pattern
• Troubleshooting - Common failures and fixes

Install

claude /plugin install https://github.com/zircote/claude-team-orchestration

Prerequisites

• Claude Code CLI 1.0.33+
• CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1 enabled
• tmux recommended for split-pane mode

Attribution

Based on the Claude Code Swarm Orchestration Skill by @kieranklaassen, decomposed into modular plugin skills.