Skill · Agent architecture
Multi-agent patterns
Design reference for multi-agent systems: when extra agents are justified, supervisor, swarm and hierarchical layouts, context isolation, consensus and failure handling.
- Author
- Muratcan Koylan (community)
- Source
- muratcankoylan/Agent-Skills-for-Context-Engineering at version
d9202c7 - Licence
- MIT
- Needs
- Works right away
- Works in
- Terminal agents such as Claude Code, Codex and Cursor
- Reviewed
- Nov 2026, skill file read in full
- Note
- Names sibling skills from the same collection (memory-systems, tool-design and others) that it does not need to work. The bundled Python file is a demo that the skill never runs.
Install only this skill
npx skills add muratcankoylan/Agent-Skills-for-Context-Engineering@d9202c76d6dc8351e648badb576a6e9b5aae1e7f --skill multi-agent-patterns -g -y
Comes in these collections
Builder
Agent Builder
Product