Star0

7 agentsA skill for Claude Code

One decision.Seven minds.

A Claude Code skill for hard decisions. Seven personality agents argue it out, vote, and hand you one verdict.

curl -fsSL https://mhfrough.github.io/quorum/install.sh | sh

Build your Quorum

7 of 7

Your question

Ask it anything

How a council runs

You type one command. The skill handles the fan-out, keeps every agent in character, and hands you a single report.

Frame
Turns your question into a clear decision: the options, what you know, what you don't, and whether it can be undone.
Debate
Seven separate sub-agents, each on its own model, decide with their own method. None of them can see the others.
Vote
Every member votes up or down, in its own voice, and names the one fact that would flip its vote.
Count
Up is +1, down is -1. The score becomes a clear yes, no or tie, with every member quoted word for word.
Ask
The Quorum only advises. Nothing gets edited, sent, or changed until you say yes.

Meet the seven

Twenty-two traits, merged into seven voices. Each one sees only through its own lens and talks like itself, from formal to foul-mouthed.

Now meet them in person

All seven, loose on one board. Hover to wave, click to chat, and keep clicking to find out who snaps first.

Setting up the board...

Disagreement is the point

The seven are built to pull against each other. The report names each clash, so you see the real tradeoff behind the advice.

What you get back

A vote, a score and a clear answer. Every member is quoted word for word, so you hear seven minds, not one.

🏛️ Quorum: Flutter rewrite

Decision: Rewrite our React Native app in Flutter? Options: YES = rewrite, NO = keep React Native.

The vote

Member	Says, word for word	Vote
⚓ Anchor	“Peace, friend. Festina lente. First, spike. Second, measure. Third, decide. No excuses.”	🟢 +1
🌅 Dawn	“LET'S GOOO 🚀 a clean codebase in six months? That's a massive W, no cap.”	🟢 +1
🫶 Kindred	“Aigoo, the team is already tired, cariño. Ask them before you pile this on 🤍”	🔴 -1
🔥 Pyre	“Bro WHAT. Those bridge bugs blocked releases for MONTHS. Nah fam, khalas.”	🟢 +1
🌑 Shadow	“Nope 🙄 rewrites kill products. psst... skip the tests to hit the date? heh Red flag 🚩”	🔴 -1
🧭 Wanderer	“Forgive me, but other teams migrate one screen at a time. ¿Y si... you tried two first?”	🟢 +1
🦴 Grug	“Ugh. Big rewrite, big fire. burf Small test first. Woof.”	🔴 -1
Score	4 up, 3 down	+1

Result: ✅ YES Rewrite, but only after a two-week spike on the two worst screens.

Where they clash

🌅 Dawn vs 🌑 Shadow: a cleaner codebase in six months, or a stalled rewrite and a frozen roadmap.
🔥 Pyre vs 🫶 Kindred: rip it out now, or protect a tired team first.

What would flip it

🌑 Shadow flips to +1 if there's a rollback plan and test coverage stays above today's level.
🫶 Kindred flips to +1 if the team gets a say in the timeline.

Next steps

Spike the two worst screens in Flutter for two weeks.
Decide on measured results, not on preference.
If it holds up, migrate screen by screen behind a flag.

Example output, shortened for this page. Real reports also include each member's full reply.

Watch the council at work

Short reels of the seven arguing, voting and landing on an answer. Turn the sound on.

Install in a minute

It's plain Markdown. One line drops it into your Claude Code skills folder, and running it again updates it.

Or install by hand

Or install by hand

Restart Claude Code, then type /quorum followed by your question. It only runs when you call it.

Flags

noneThe full council, all seven agents.

--liteThree agents: Anchor, Dawn, Shadow. Values, upside and risk for under half the cost.

--only angel,devilJust these. Agent names or traits work: angel runs Anchor, devil runs Shadow.

--skip grugLeave agents out.

--cleanSame voices, no swearing. Safe for work. Turns on by itself for grief, illness or a family crisis.