agentleFS
khgouldykhgouldy/skills/skills/productivity/cross-critique/SKILL.md

cross-critique

Skill1 starsChanged 16 days agoMITView on GitHub

Use when a decision is high-stakes or genuinely uncertain and you want it stress-tested before committing — dispatch independent critiques of a proposed choice, then synthesize. Triggers include "pressure-test this decision", "what am I missing", "red-team this", choosing between architectures, or any irreversible call worth a second and third opinion.

2 other projects use a version of this.

Adopted by

Most recently taken up by

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
  1. Distinctive:

    Dispatch independent critics — give each a distinct lens: Don't ask three agents the same question.

    line 25
  2. Distinctive:

    Run them independently so they don't anchor on each other.

    line 36
  3. Distinctive:

    Synthesize — don't just average: Read the critiques and decide what actually changes your mind.

    line 42

≈ “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

Cross-Critique

Sharpen a decision by attacking it from several independent angles before you commit, then synthesize what survives. A single line of reasoning has blind spots; multiple independent critiques expose the ones you can't see from inside your own argument.

When it's worth it

For reversible, low-stakes calls, just decide. Reach for cross-critique when the decision is expensive to undo or genuinely uncertain: architecture choices, data-model commitments, anything where being wrong costs days.

How to run it

  1. State the proposal crisply. One clear statement of the decision and the reasoning behind it. Vague proposals get vague critiques.
  2. Dispatch independent critics — give each a distinct lens. Don't ask three agents the same question; redundant critics find redundant problems. Assign different angles, e.g.:

    • Correctness / logic — where does the reasoning not hold?
    • Risk / failure modes — how does this go wrong in production, at scale, at 3am?
    • Cost / simplicity — is there a materially cheaper or simpler option that gets 90% of the value?
    • The opposite — argue the strongest case for the alternative we're rejecting.

    Run them independently so they don't anchor on each other.

  3. Demand specifics, not vibes. A critique that says "seems risky" is noise. Each critic should name a concrete failure, a concrete cheaper alternative, or a concrete flawed assumption.
  4. Synthesize — don't just average. Read the critiques and decide what actually changes your mind. Some objections are decisive; some are theoretical. The output is a sharper decision (possibly the original one, now better-justified), not a tally of votes.

Output

A short writeup: the decision, the objections that mattered, how they changed (or didn't change) the call, and the residual risks you're knowingly accepting. That record is itself valuable — it's the "why" a future handoff or ADR needs.


Inspired by Warp's cross-critique skill (https://github.com/warpdotdev/common-skills). Prose is our own.

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.