Skip to content
Get packs

Harness agent-team designer

Designs multi-agent teams and per-agent skills for a project, with workflow and model choices.

Source Download skill
Repository
revfactory/harness
Stars
9,305
Licence
Apache-2.0
Files
10 files

What it does

When installed, the agent can plan a set of specialized AI agents for a project: it defines each agent's role in a file, chooses how they run together (scripted workflow, persistent collaborating agents, or one-off sub-agents), picks a model per task, and writes the skills those agents follow. It also checks an existing setup for mismatches and records invocation notes in CLAUDE.md so the setup can be reused in later sessions.

Great for

  • Developers building an automated multi-agent setup for a codebase or project
  • Teams that want to extend or audit an existing agent/skill configuration
  • Users who want a structured plan before creating agent definitions

Skip it if

  • Simple one-off tasks that a single prompt can handle, since the setup process adds overhead
  • Users who do not understand agent configuration files and would accept generated files without review, because generated agents and workflows can run commands and spend significant resources

Try asking

  • “Set up an agent team for reviewing and refactoring our Node.js project”
  • “Design a harness for a data-cleaning pipeline with a validator agent”
  • “Audit our existing .claude/agents and skills folders for mismatches with CLAUDE.md”
Step-by-step guide: how to use Harness agent-team designerInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 10 original files, the Apache-2.0 licence and an attribution note.

    harness.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip harness.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip harness.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Review the proposed team structure and model choices before approving creation of files.
  • Ask explicitly for a limited number of agents first; broad 'thorough' requests can trigger large parallel runs.
  • Keep the skill's generated outputs in the language you use; the skill is written in Korean but adapts to your language.

What's inside

  • references/execution-modes.md
  • references/model-selection-guide.md
  • references/orchestrator-template.md
  • references/qa-agent-guide.md
  • references/skill-testing-guide.md
  • references/skill-writing-guide.md
  • references/team-examples.md
  • references/team-patterns.md
  • references/workflow-recipes.md
  • SKILL.md

Redistributed unmodified from revfactory/harness under the Apache-2.0 licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.