67
Skill11 starsChanged 3 months agoMITView on GitHub
Runs the task through six agents in parallel, each locked to a different strategy, then a seventh judge ranks the six attempts and synthesizes a final answer that steals the strongest part of each; invoke via /67 for high-stakes or open-ended work where a single attempt would leave quality on the table.
In this repository4 other files it gives its agents.
- aura-farmingSkill
- china-maxingSkill
- crash-outSkill
- larpSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 3 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 13
Not for trivial tasks. Six agents on a one-liner is theater.
- Distinctive: line 18
Write a single task brief every agent receives verbatim: the task, the deliverable format, and the constraints.
- Distinctive: line 34
If no sub-agent tool is available, run six sequential passes with a hard reset between each: "discard the previous approach entirely.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
67
One attempt is a sample size of one. This runs six, each forced down a different road, then a seventh agent judges the field, ranks it, and builds the final from the best parts. You ship the strongest version that exists, not the first one that came out.
When to use
- High-stakes deliverables: the pitch, the architecture, the pricing page, the cold email.
- Open-ended problems with many valid approaches and no obvious winner.
- Not for trivial tasks. Six agents on a one-liner is theater.
Behavior
1. Frame once
Write a single task brief every agent receives verbatim: the task, the deliverable format, and the constraints. Tight enough that all six outputs are directly comparable.
2. Fan out six agents in parallel
Spawn all six in one message (Agent/Task tool). Each gets the same brief plus ONE strategy lock it must follow:
| # | Lens | Instruction |
| 1 | Minimalist | Smallest thing that fully works. Cut everything optional. |
| 2 | Maximalist | Most complete version. Cover every edge and every objection. |
| 3 | First principles | Ignore convention. Derive the answer from scratch. |
| 4 | Prior art | Find how the best existing versions do it and adapt the strongest pattern. |
| 5 | Adversarial | Design against failure. Ask what breaks each choice, keep what survives. |
| 6 | Wildcard | One unreasonable creative bet, executed seriously. |
Each agent must return a complete, standalone answer. No hedging, no "it depends" without a decision.
If no sub-agent tool is available, run six sequential passes with a hard reset between each: "discard the previous approach entirely; you are now agent N with this lens."
3. Agent seven judges
The judge never writes its own answer first. It scores each attempt 1 to 10 on correctness, completeness, and taste; ranks all six with a one-line verdict each; and names each attempt's single strongest element, including the losers'.
4. Synthesize
Build the final on the winner's skeleton, then graft in every strongest element worth stealing. Note what was taken from whom.
Output
The leaderboard (rank, score, one-line verdict per agent), then the synthesized final, then a short "stolen parts" ledger.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.