multi-agent-patterns

Name: multi-agent-patterns
Author: muratcankoylan

90installs1.4kforks

How to install multi-agent-patterns?

muratcankoylan/agent-skills-for-context-engineering/multi-agent-patterns

$npx -y skills add muratcankoylan/Agent-Skills-for-Context-Engineering --skill multi-agent-patterns

Installs into the current project.

Summary

This skill should be used when designing multi-agent systems that need context isolation, supervisor or swarm coordination, explicit handoffs, parallel execution, or a decision on whether multiple agents are justified.

Files · 1

View on GitHub

SKILL.md

1	# Multi-Agent Architecture Patterns
2
3	Multi-agent architectures distribute work across multiple language model instances, each with its own context window. When designed well, this distribution enables capabilities beyond single-agent limits. When designed poorly, it introduces coordination overhead that negates benefits. The critical insight is that sub-agents exist primarily to isolate context, not to anthropomorphize role division.
4
5	## When to Activate
6
7	Activate this skill when:
8	- Single-agent context limits constrain task complexity
9	- Tasks decompose naturally into parallel subtasks
10	- Different subtasks require different tool sets or system prompts
11	- Building systems that must handle multiple domains simultaneously
12	- Scaling agent capabilities beyond single-context limits
13	- Designing production agent systems with multiple specialized components
14
15	Do not activate this skill for adjacent work owned by other skills:
16	- Deciding task-model fit, pipeline shape, or project-level cost before topology is known: `project-development`.
17	- Designing hosted sandboxes, warm pools, remote sessions, or background runtime infrastructure: `hosted-agents`.
18	- Sharing orchestrator state through KV-cache compaction in controlled runtimes: `latent-briefing`.
19	- Designing the tools each agent exposes: `tool-design`.
20
21	## Core Concepts
22
23	Use multi-agent patterns when a single agent's context window cannot hold all task-relevant information. Context isolation is the primary benefit — each agent operates in a clean context without accumulated noise from other subtasks, preventing the telephone game problem where information degrades through repeated summarization.
24
25	Choose among three dominant patterns based on coordination needs, not organizational metaphor:
26
27	- Supervisor/orchestrator — Use for centralized control when tasks have clear decomposition and human oversight matters. A single coordinator delegates to specialists and synthesizes results.
28	- Peer-to-peer/swarm — Use for flexible exploration when rigid planning is counterproductive. Any agent can transfer control to any other through explicit handoff mechanisms.
29	- Hierarchical — Use for large-scale projects with layered abstraction (strategy, planning, execution). Each layer operates at a different level of detail with its own context structure.
30
31	Design every multi-agent system around explicit coordination protocols, consensus mechanisms that resist sycophancy, and failure handling that prevents error propagation cascades.
32
33	## Detailed Topics
34
35	### Why Multi-Agent Architectures
36
37	The Context Bottleneck
38	Reach for multi-agent architectures when a single agent's context fills with accumulated history, retrieved documents, and tool outputs to the point where performance degrades. Recognize three degradation signals: the lost-in-middle effect (attention weakens for mid-context content), attention scarcity (too many competing items), and context poisoning (irrelevant content displaces useful content).
39
40	Partition work across multiple context windows so each agent operates in a clean context focused on its subtask. Aggregate results at a coordination layer without any single context bearing the full burden.
41
42	The Token Economics Reality
43	Budget for substantially higher token costs. Production data shows multi-agent systems can cost far more tokens than single-agent chat (claim-multi-agent-token-multiplier):
44
45	\| Architecture \| Token Multiplier \| Use Case \|
46	\|--------------\|------------------\|----------\|
47	\| Single agent chat \| Baseline \| Simple queries \|
48	\| Single agent with tools \| Higher than baseline \| Tool-using tasks \|
49	\| Multi-agent system \| Much higher than baseline \| Complex research/coordination \|
50
51	Browsing-agent evaluation research suggests token usage, tool calls, and model choice dominate performance variance (claim-evaluation-browsecomp-variance). This supports measuring multi-agent setups against single-agent baselines instead of assuming extra agents help.
52
53	Prioritize model selection alongside architecture design — upgrading to better models often provides larger performance gains than doubling token budgets. BrowseComp data shows that model quality improvements frequently outperform raw token increases. Treat model selection and multi-agent architecture as complementary strategies.
54
55	The Parallelization Argument
56	Assign parallelizable subtasks to dedicated agents with fresh contexts rather than processing them sequentially in a single agent. A research task requiring searches across multiple independent sources, analysis of different documents, or comparison of competing approaches benefits from parallel execution. Total real-world time approaches the duration of the longest subtask rather than

People who liked this also viewed

6 picks

Type

multi-agent-patterns

90installs1.4kforks

View on GitHub

How to install multi-agent-patterns?

muratcankoylan/agent-skills-for-context-engineering/multi-agent-patterns

$npx -y skills add muratcankoylan/Agent-Skills-for-Context-Engineering --skill multi-agent-patterns

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Multi-Agent Architecture Patterns
2
3	Multi-agent architectures distribute work across multiple language model instances, each with its own context window. When designed well, this distribution enables capabilities beyond single-agent limits. When designed poorly, it introduces coordination overhead that negates benefits. The critical insight is that sub-agents exist primarily to isolate context, not to anthropomorphize role division.
4
5	## When to Activate
6
7	Activate this skill when:
8	- Single-agent context limits constrain task complexity
9	- Tasks decompose naturally into parallel subtasks
10	- Different subtasks require different tool sets or system prompts
11	- Building systems that must handle multiple domains simultaneously
12	- Scaling agent capabilities beyond single-context limits
13	- Designing production agent systems with multiple specialized components
14
15	Do not activate this skill for adjacent work owned by other skills:
16	- Deciding task-model fit, pipeline shape, or project-level cost before topology is known: `project-development`.
17	- Designing hosted sandboxes, warm pools, remote sessions, or background runtime infrastructure: `hosted-agents`.
18	- Sharing orchestrator state through KV-cache compaction in controlled runtimes: `latent-briefing`.
19	- Designing the tools each agent exposes: `tool-design`.
20
21	## Core Concepts
22
23	Use multi-agent patterns when a single agent's context window cannot hold all task-relevant information. Context isolation is the primary benefit — each agent operates in a clean context without accumulated noise from other subtasks, preventing the telephone game problem where information degrades through repeated summarization.
24
25	Choose among three dominant patterns based on coordination needs, not organizational metaphor:
26
27	- Supervisor/orchestrator — Use for centralized control when tasks have clear decomposition and human oversight matters. A single coordinator delegates to specialists and synthesizes results.
28	- Peer-to-peer/swarm — Use for flexible exploration when rigid planning is counterproductive. Any agent can transfer control to any other through explicit handoff mechanisms.
29	- Hierarchical — Use for large-scale projects with layered abstraction (strategy, planning, execution). Each layer operates at a different level of detail with its own context structure.
30
31	Design every multi-agent system around explicit coordination protocols, consensus mechanisms that resist sycophancy, and failure handling that prevents error propagation cascades.
32
33	## Detailed Topics
34
35	### Why Multi-Agent Architectures
36
37	The Context Bottleneck
38	Reach for multi-agent architectures when a single agent's context fills with accumulated history, retrieved documents, and tool outputs to the point where performance degrades. Recognize three degradation signals: the lost-in-middle effect (attention weakens for mid-context content), attention scarcity (too many competing items), and context poisoning (irrelevant content displaces useful content).
39
40	Partition work across multiple context windows so each agent operates in a clean context focused on its subtask. Aggregate results at a coordination layer without any single context bearing the full burden.
41
42	The Token Economics Reality
43	Budget for substantially higher token costs. Production data shows multi-agent systems can cost far more tokens than single-agent chat (claim-multi-agent-token-multiplier):
44
45	\| Architecture \| Token Multiplier \| Use Case \|
46	\|--------------\|------------------\|----------\|
47	\| Single agent chat \| Baseline \| Simple queries \|
48	\| Single agent with tools \| Higher than baseline \| Tool-using tasks \|
49	\| Multi-agent system \| Much higher than baseline \| Complex research/coordination \|
50
51	Browsing-agent evaluation research suggests token usage, tool calls, and model choice dominate performance variance (claim-evaluation-browsecomp-variance). This supports measuring multi-agent setups against single-agent baselines instead of assuming extra agents help.
52
53	Prioritize model selection alongside architecture design — upgrading to better models often provides larger performance gains than doubling token budgets. BrowseComp data shows that model quality improvements frequently outperform raw token increases. Treat model selection and multi-agent architecture as complementary strategies.
54
55	The Parallelization Argument
56	Assign parallelizable subtasks to dedicated agents with fresh contexts rather than processing them sequentially in a single agent. A research task requiring searches across multiple independent sources, analysis of different documents, or comparison of competing approaches benefits from parallel execution. Total real-world time approaches the duration of the longest subtask rather than

People who liked this also viewed

6 picks

Type