agentleFS
rubiconrubicon/ai-skills/skills/showdown/SKILL.md

showdown

Skill1 starsChanged 3 months agoMITView on GitHub

Use when the user knows what they want to build or change but not how to approach it, and wants the options compared before anyone plans or codes. Triggers on "how should I approach this", "which way should I go", "compare my options", "what are the trade-offs", "help me decide between", "is X or Y better here", or when the user is weighing architectures, libraries, or designs for a known goal.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 5 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    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:

    Put two or three genuinely different approaches head to head on the same terms, pick one, and stop.

    line 15
  2. Distinctive:

    Ask a question only when a missing fact would change which options exist or which one wins.

    line 21
  3. Distinctive:

    Every cell is specific to this repo and this goal.

    line 38
  4. Distinctive:

    Before you build: only the open questions whose answer would change the pick, numbered.

    line 42
  5. Distinctive:

    If you find a bug or a blocker in the existing code while investigating, state it in one sentence before the Goal line, and treat fixing it …

    line 46

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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

Showdown

Put two or three genuinely different approaches head to head on the same terms, pick one, and stop. The output is a decision, not a plan and not code. Once the user picks, planning or building is a separate step.

1. Investigate first

Read the code the decision touches: the relevant modules, the existing patterns, the data shapes, and the constraints. Every option and every cell in the comparison must be grounded in what the repo actually does. Make sure you understand what the user is trying to achieve and why.

Ask a question only when a missing fact would change which options exist or which one wins. Ask it in plain text, numbered, at most three, and wait.

2. Choose the contenders

Two or three options. Each one must be an approach a competent engineer would choose in some real situation, and you must be able to name that situation. If you cannot, the option is a straw man: replace it with a real alternative or drop to two.

Include the option that fully delivers the goal even when it is the hardest. Include the cheapest option that plausibly works, so the user sees what the extra effort buys.

3. Answer in this shape

Goal: one sentence restating what the user is trying to achieve, in their terms.

The contenders: one table, one row per option, these columns in this order:

OptionWhat it involvesHow fully it meets the goalFit with this codebaseEffort and riskRight choice when

Every cell is specific to this repo and this goal. "Right choice when" names a concrete situation in which that option wins, stated on its own terms, not "it depends". Whether that situation applies to this user is argued in the Pick.

Pick: one paragraph. Which option, and why, argued from the goal. Effort and risk are reasons only when two options meet the goal about equally well. Then one sentence that starts "Pick <other option> instead if" and names an observable condition that would flip the decision.

Before you build: only the open questions whose answer would change the pick, numbered. Omit this part when there are none.

The answer ends there. Anything you could not check while investigating, such as tests you could not run, goes in one sentence inside the opening bug line or a Before you build question, not after them.

If you find a bug or a blocker in the existing code while investigating, state it in one sentence before the Goal line, and treat fixing it as a constraint every option must meet.

Example

Goal: let support staff resend a failed order-confirmation email without asking engineering.

OptionWhat it involvesHow fully it meets the goalFit with this codebaseEffort and riskRight choice when
Admin button that re-enqueues the existing SendConfirmation jobOne endpoint and one button in the existing admin panelFully, for single ordersReuses the job queue and audit log already in admin/Low; double-sends possible without an idempotency keyResends are occasional and one at a time
Bulk resend screen with filtersA filtered order list, select-all, and a batch jobFully, including outage recoveryNew batch job type; the admin panel has no bulk actions yetMedium; a wrong filter can email thousandsOutages produce hundreds of failures at once
Self-serve "resend my email" link for customersA public endpoint with rate limitingPartly; support still handles customers who never got the first emailNeeds a public route, which the app has none ofMedium; new abuse surfaceMost failures are spam-folder issues, not send failures

Pick: the admin button. It fully meets the goal as stated, reuses the queue and audit log, and adding an idempotency key closes its one real risk. Pick the bulk screen instead if the failure log shows more than a few dozen failures at a time.

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.