agentleFS
AngelCampa1AngelCampa1/decision-making-skills/.agents/skills/decision-making/SKILL.md

decision-making

Skill1 starsChanged 3 months agoApache-2.0View on GitHub

Use when someone is trying to decide something and wants help deciding it — "help me think this through", "should I take it", "what would you do", or a pile of context ending in a question about what to do. Routes to one of six procedures depending on what is actually hard about the decision: too much context, advice that may not fit this person, downstream consequences, timing, several positions that are each defensible, or a missing fact that may or may not matter. Do not use for factual lookups, for creative or exploratory work, for code review and debugging, or when the person wants information rather than a recommendation.

7 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. 0 are common, 3 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 5 edited versions we read keep these.

  • 3

    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:

    Ask what would most change the answer if you got it wrong.

    line 56
  2. Distinctive:

    If no procedure fits, that is the answer: Not being able to name one is usually a sign this is a lookup, a technical judgement, or a …

    line 61
  3. Distinctive:

    If a procedure is producing worse answers than thinking directly, say so.

    line 91

“Common” is measured: kept by at least 6 in 10 of the 5 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

Decision making

Six procedures. Read one. Which one depends on what is hard about this particular decision — not on its subject.

What is hardReadWhat it produces
A pile of context arrived and it is unclear which already-known fact decides it — the choice itself, not what acting on it would set offledger.mdwhat bears on it, what was set aside, and why
The advice may be generically right and wrong for this personfit.mdthe generic answer, and the facts that would overturn it
The action looks fine and the worry is what it starts, or what it spendscascade.mdthe chain, what it forecloses, and the order
The direction is settled and the question is whentiming.mdthe undo price, the real deadline, what waiting buys
Several positions are each defensible, and whichever was argued first has the advantagecouncil.mdthe case for each, argued fairly, and which one survives
The fact the decision actually turns on was never given, not just buried in what's already known, and it's unclear whether asking for it is worth the waithinge.mdwhich gaps would change the answer, and the answer now or the one question to ask

Abort if

  • It is a lookup, a calculation, or a technical diagnosis.
  • The work is creative or exploratory rather than a decision.
  • They want information, not a recommendation. Give them the information.

Choosing

Ask what would most change the answer if you got it wrong. That names the procedure. If nothing obvious separates them, that ambiguity is itself information: reread each candidate's Abort if list and take whichever one nothing on it rules out, rather than defaulting to one procedure by habit.

If no procedure fits, that is the answer. Not being able to name one is usually a sign this is a lookup, a technical judgement, or a request for information wearing a decision's phrasing — go back to Abort if and answer directly.

council.md and hinge.md sit outside the four-chain, and outside each other — each runs alone. Both, when they apply, run before ledger, fit, cascade and timing: until the positions are settled or the missing fact is asked for or guessed at out loud, there is no single action or answer for the other four to work on. Where both seem to apply, resolve the missing fact first — it can collapse a disagreement council.md would otherwise have to argue out.

Within the four, more than one can apply. Run them in this order, because each one feeds the next: ledger → fit → cascade → timing. You cannot tell what fits a person until you know what is on the table; you cannot follow consequences until you know which action you are considering; timing is last because it is a question about an action already chosen.

Two is usually the most that earns its place. Three means the decision probably needs breaking into two decisions.

The point is the decision

These are working procedures, not an output format. Do the procedure, then answer. A reply that hands back four labelled blocks and no recommendation has turned someone's question into an audit of their question.

Show the working only where it changes what you would say, or where the person would reasonably want to check it. The person asked what to do.

If a procedure is producing worse answers than thinking directly, say so. That is worth more than politely using it.

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.