Skip to content
repository radar
Agents & Orchestration

🏭harness

ACTIVE BREAKOUT

revfactory/harness · homepage ↗

A meta-skill that designs domain-specific agent teams, defines specialized agents, and generates the skills they use.

⭐ Popular: 9k stars, gaining about 83 a week

View on GitHub ↗

repo profile

vintage 2026 launched this year
language HTML
license Apache-2.0

momentum

total stars 9k
stars added last week +83
commits / week 0 steady
issues closed 73% ~13d to close, median
contributors 6
release cadence n/a
last activity 22d ago

durability

backing community / independent
openness permissive
bus factor 1 solo
top-author share 78% 6 mo

bus factor = how many people it takes to cover more than half the commits (6 months). 1 is a solo project; higher means the work is spread across a team. top-author share is the single busiest author's slice of those commits.

since we covered it

monthly average + 1k/mo (+24%/mo) · + 3k total since PR#35

why it's a big deal

  • It automates the setup step for multi-agent Claude Code projects, turning a plain domain description into a working agent team plus the skills those agents use.
  • It targets Claude Code users who want structured multi-agent workflows but do not want to hand-write agent definitions and skill files themselves.
  • It constrains the design to six named team patterns, so the generated structure follows a known shape rather than an ad hoc one.

under the hood

  • The core is a single meta-skill defined in skills/harness/SKILL.md that runs a six-phase workflow covering domain analysis, team architecture design, agent generation, skill generation, orchestration, and validation.
  • It writes markdown outputs into the project's.claude directory, with agent definitions under.claude/agents and skills under.claude/skills, and picks from six patterns including pipeline, fan-out/fan-in, expert pool, producer-reviewer, supervisor, and hierarchical delegation.
  • It installs via the Claude Code marketplace or by copying the skills directory, requires the experimental Agent Teams flag CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS set to 1, and is licensed Apache 2.0 with the repo counted as HTML because of its published docs site.

Radar summary, generated from the project's public sources

star history

PR#35 · 6k9k now Mar 2026Aug 2026
  1. PR#35 6k 2026-06-03
  2. now 9k + 3k since first covered

curve is sampled from GitHub's star history, plus our own daily readings since we covered it; the dashed stretch is before we first covered it, the solid line since. figures at coverage are the numbers we printed then (approx.), current count is live.

understory

Better known than its recent output, coasting a little on attention.

-49 understory score output 24 · clout 73
Aug 2025 Jul 2026
  • output, commits & releases
  • clout, star velocity

output = commits & releases; clout = star velocity, both 0 to 100 monthly indices; the gap where output runs above clout is the understory. The understory →

covered in

  • PR#35 2026-06-03 below the radar

    A factory that builds whole agent teams from one sentence

similar projects

compare these →
  • 🌊 Ruflo

    TypeScript · 7.7× the stars

    🌊 The original agent meta-harness. Deploy intelligent multi-player swarms, coordinate autonomous workflows, and build conversational AI systems. Features adaptive memory, self-learning intelligence, RAG integration, and native Claude Code / Codex / Hermes and many more Integrated

    68k ACTIVE
  • ☤ hermes-agent

    Python · 26× the stars

    The agent that grows with you

    231k ACTIVE
  • 🧩 cc switch

    Rust · 15× the stars

    A cross-platform desktop All-in-One assistant for Claude Code, Codex, OpenCode, OpenClaw, Grok Build & Hermes Agent. Only official website: ccswitch.io

    127k ACTIVE

comments

Sign in with GitHub to add your blip on harness.

loading comments…