Agents run as separate sub-processes with an isolated context budget. Use one when a task needs its own working memory or a narrower tool set than the main loop.

Overview

What an agent is and how it differs from a skill.

Choosing an agent

Matching a task to the right specialist instead of the generic one.

Multi-agent patterns

Fan-out, pipelines, and adversarial verification.

Writing agents

Frontmatter, tool scoping, and skill wiring.

/ork:memory

Search, load, sync, and visualize your knowledge graph across sessions.

Agents Overview

37 specialized agents -- how they activate, what they know, and when to use each.

Edit on GitHub

Last updated on