agentleFS
Sign inStart free
malinskibeniaminmalinskibeniamin/skills/steelman/SKILL.md

steelman

Skill4 starsChanged 11 days agoMITView on GitHub

Argue the strongest evidence-backed case against a premise. Use when user says steelman, asks for pushback or second opinion, or when a high-stakes decision depends on an uncertain assumption.

13 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

2 rules. 0 are common, 2 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 10 edited versions we read keep these.

  • 2

    Distinctive

    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:

    Evidence first: search symbols/patterns, read cited files, run cheap checks, consult current docs.

    line 16
  2. Distinctive:

    Never ask "are you sure?", play ungrounded devil's advocate, gate the user, or steelman every turn.

    line 23

“Common” is measured: kept by at least 6 in 10 of the 10 edited versions of this file we could read. “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

Counter sycophancy with evidence. Skip preferences/goals/scope, trivial actions, proven claims, and implementation unless security, data loss, or irreversibility warrants pushback.

Procedure

  1. Claim: restate once and classify:
    • factual -> verify;
    • causal -> test mechanism;
    • architectural -> inspect actual usage;
    • merge premise ("this PR should merge") -> argue from the product: verdict, lie-detector evidence, and review findings;
    • preference/goal/scope -> return noise; user's call.
  2. Evidence first: search symbols/patterns, read cited files, run cheap checks, consult current docs. Never argue from generic smells.
  3. Opposite: 2-4 bullets with file:line or command output. State what must be true for the claim to fail, supporting repo evidence, missed failure mode, and contradictory precedent/history.
  4. Verdict:
    • Confirmed: evidence supports user; proceed.
    • Contradicted: show evidence and let user override/revise; never block.
    • Mixed: separate supported/unsupported parts.

Never ask "are you sure?", play ungrounded devil's advocate, gate the user, or steelman every turn. Reserve for explicit invitation or high stakes.

[ETHOS: User fallible. Verify before act. Surface evidence, not doubt.]

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.

Check your setup, free