Fan Out

SkillAI & models

Use single-model sub-agents for independent angles. Use for "in parallel", "multiple angles", "compare approaches", and broad analysis.

Available today. Use it from your connected AI after setup.

Connect ahel once, and every AI you use reads what you have installed.

Then ask your AI: use the Fan Out skill

What this skill tells your AI

The instructions your AI receives, as published by uss-parks/lamprey-harness in skills/fan-out/SKILL.md and read by ahel’s review.

Use this skill when the task decomposes into independent reasoning passes that can be checked separately.

Prefer multi_agent_run when the branches do not need tools: for example planner plus reader, reviewer plus verifier, or multiple architecture comparisons over supplied context. Keep each sub-agent prompt narrow. Provide bounded context, explicit output format, and only the facts needed for that role.

Use the active chat model for every sub-agent. Do not use fan-out to create a visible multi-model conversation or to bypass tool approval. Sub-agents must not call tools; tool use stays in the main run after their outputs return.

Good role sets: planner for decomposition, reader for fact extraction, verifier for checking claims, reviewer for code risk, and coworker for a second implementation angle.

After fan-out, synthesize the results yourself. Name disagreements and choose one path.

Stop when the independent angles converge or when the disagreement identifies the next concrete read or verification step.

Signals

GitHub stars
62
Forks
2
Last commit
Sep 2026
Advanced
Catalog kind
skill
Gateway key
fan-out-uss-parks
Source
github.com/uss-parks/lamprey-harness