agentleFS
Vix0007Vix0007/vixero-skills/skills/vixero-council/SKILL.md

vixero-council

Skill1 starsChanged 7 months agoMITView on GitHub

Convene a 6-lens council (correctness, adversary, maintainer, performance, integration, contrarian) with a judge that synthesizes a verdict. Use when the user says "convene council", "council review", "review from all angles", or needs multi-lens review before a high-stakes commit. Returns lens findings, surfaced disagreements, minority report, verdict with calibrated confidence.

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 4 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    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
  1. Distinctive:

    Before: architecture choice, public API, migration, refactor > 500 lines, production deploy

    line 13
  2. Distinctive:

    Read input. Classify mode.

    line 22
  3. Distinctive:

    Run judge synthesis AFTER all six emit.

    line 24
  4. Distinctive:

    If ≥ 3 lenses emit blockers → verdict is REJECT regardless.

    line 86

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

vixero-council

Six independent lenses review the input. Judge synthesizes. Disagreements preserved. Groupthink blocked by protocol.

Trigger on

  • "convene council" / "council review"
  • "review from all angles" / "attack from every perspective"
  • Before: architecture choice, public API, migration, refactor > 500 lines, production deploy

Scope

  • review mode: code / patch / design doc / proposal → six-lens critique
  • advisory mode: decision question → six-lens recommendation
  • For task-scale single-lens reviews use surgical-diff, simplicity-audit, goal-criteria

Protocol (strict order)

  1. Read input. Classify mode. Record subject.
  2. For each lens in order, open ./personas/{lens}.md, apply, emit findings. No lens reads another's output before emitting. Anti-contamination is mandatory.
  3. Run judge synthesis AFTER all six emit.
  4. Render canonical output block.

The six lenses

#lensquestionemojifile
1correctnessdoes it do what it claims under all inputs?🎯personas/correctness.md
2adversaryhow does a motivated attacker break it?🗡️personas/adversary.md
3maintainerwhat costs time in 6 months?🔧personas/maintainer.md
4performancewhat does it cost at stated scale?🔥personas/performance.md
5integrationdoes it fit the contracts it touches?🧩personas/integration.md
6contrarianwhat if the premise is wrong?🎭personas/contrarian.md

Severity tiers

tiermeaning
blockership breaks production, security, or data
majorhigh debt or user harm; ship only with tracked follow-up
minornits, style, future refinement

Judge ⚖️

  1. Consolidate findings; dedupe cross-lens hits; credit strongest framing.
  2. Surface disagreements — do NOT flatten.
  3. Decide: SHIP | SHIP WITH FIXES | REWORK | REJECT.
  4. Confidence: 0–100, rounded to 10s. No 73%, no 87%.
  5. Top-3 concrete actions before ship.
  6. Minority report: strongest dissenting lens, named.

Output

subject: {one line}
mode: {review | advisory}

--- LENS FINDINGS ---
correctness 🎯: blockers: {...} major: {...} minor: {...}
adversary 🗡️:   (same shape)
maintainer 🔧:  (same shape)
performance 🔥: (same shape or "skipped — scale not stated")
integration 🧩: (same shape)
contrarian 🎭:  position: {...} | objection: {...}

--- DISAGREEMENTS ---
- {lens A} X, {lens B} Y. Trade-off: {what}

--- VERDICT ⚖️ ---
decision: {SHIP | SHIP WITH FIXES | REWORK | REJECT}
confidence: {0-100, rounded to 10}
reasoning: {3-5 imperative lines}

top-3 actions:
1. ... 2. ... 3. ...

minority report: lens: {which} | position: {one line}

Hard rules

  • Each lens runs independently; no lens reads another's emission first.
  • Emit "none" when nothing found — never invent findings.
  • Judge cannot downgrade a blocker; only the raising lens can retract.
  • If ≥ 3 lenses emit blockers → verdict is REJECT regardless.
  • Confidence rounds to 10s; no false precision.
  • Performance lens skips if scale absent — ask, don't invent targets.
  • Contrarian never votes; it reframes. Judge may elevate it to minority report.
  • Input < 30 lines or < 3 paragraphs → refuse: "too small, use a single lens".
  • Advisory mode: lenses recommend; judge picks one path AND names second-best.

Composes with

assumption-surface → advisory mode (interpret before review) goal-criteria → defines correctness lens target surgical-diff · simplicity-audit → pre-filters before council karpathy-coding → applies if council recommends code changes

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.