skills.rest
luicabref97
luicabref97COMMUNITY
Explore all skills by this author →
GitHub

sadd:multi-agent-patterns

Design multi-agent architectures for complex reasoning and coordination tasks.

Updated Apr 6, 2026
Search Tags:#multi-agent#orchestration#consensus#coordination#context-isolation#task-decomposition#file-system-memory
One-click install
npx skills add https://github.com/luicabref97/sushi-jungle-web --skill sadd-multi-agent-patterns-luicabref97
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: sadd:multi-agent-patterns
Source: https://github.com/luicabref97/sushi-jungle-web/tree/main/.agents/skills/sadd-multi-agent-patterns
Command: npx skills add https://github.com/luicabref97/sushi-jungle-web --skill sadd-multi-agent-patterns-luicabref97

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill addresses the limitations of single-agent contexts by providing patterns and practices to decompose, parallelize, and coordinate complex tasks across multiple focused agents so that reasoning, tool use, and state do not overload a single context window.

Core Features & Use Cases

  • Architectural Patterns: Supervisor/orchestrator, peer-to-peer/swarm, and hierarchical patterns for different coordination needs.
  • Context Isolation: Techniques for instruction passing, file-system shared memory, and controlled context delegation to avoid context bloat.
  • Consensus & Safety: Voting, weighted contributions, and debate/critique protocols to reduce sycophancy and error propagation.
  • Memory & Persistence: File-based working, session, and long-term memory patterns including handoff files, progress tracking, and temporal reasoning recommendations.
  • Use Case Examples: Parallel research and analysis, multi-specialist code review (security, performance, style), and layered project planning and execution.

Quick Start

Activate the sadd:multi-agent-patterns skill and ask it to decompose my complex task into specialist subagents, coordinate parallel execution via file-based memory, and aggregate validated results.

Frequently Asked Questions about sadd:multi-agent-patterns

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

FAQPage Schema
How do I coordinate multiple agents when a complex task exceeds a single context window?▼

Multi-agent coordination decomposes complex tasks into focused subagents using architectural patterns like supervisor, peer-to-peer, or hierarchical orchestration to distribute reasoning and prevent context overload.

What is file-system based shared memory for multi-agent orchestration?▼

File-system based shared memory is a context isolation technique that uses handoff files and progress tracking to pass instructions and persist state between agents without causing context bloat.

How do I decompose a complex workflow into specialized subagents?▼

Task decomposition breaks down complex workflows into focused subagents by applying explicit coordination protocols, isolating context, and aggregating validated outputs to manage parallel research or modular code review.

Can multi-agent consensus mechanisms mitigate sycophancy and error propagation?▼

Consensus mechanisms like voting, weighted contributions, and debate protocols mitigate sycophancy and error propagation by validating outputs across multiple agent contexts before final aggregation.

When should I use hierarchical multi-agent patterns instead of a single agent?▼

Hierarchical multi-agent patterns are necessary for complex layered planning and execution, specialized tool orchestration, or multi-specialist code review that exceeds single-agent context limits and requires structured task delegation.

Does multi-agent context isolation require specific frameworks to manage shared memory?▼

Context isolation uses file-system based working, session, and long-term memory patterns with handoff files to delegate context and avoid bloat, without requiring external framework dependencies.

Related Skills

Explore More Software Engineering Skills→
Tyler-R-KendrickTyler-R-Kendrick
1

nemo-rl-session-memory

Create timestamped session directories with state, timeline, and handoff documentation.

Community
Intermediate
Hameem1Hameem1

session-wrapup

Completes project wrap-up including linting, validation and commit for a session.

Community
Advanced
scooter-lacroixscooter-lacroix
8

sub-agents

Create and configure specialized AI sub-agents with custom prompts, tools, and models.

Community
Intermediate
chrono-metachrono-meta
7

context-bridge-dispatch

Coordinate parallel AI coding agents with context injection and task boundaries.

Community
Intermediate