council

Overview of council skill

council is a decision-support skill for moments when there are several plausible answers and you need structured disagreement before you commit. It is best for product calls, scope tradeoffs, rollout choices, and “ship or hold?” questions where a normal single-response prompt tends to overfit one viewpoint. If you want the council skill to help you choose, not just brainstorm, it is a good fit.

What council is for

The council for Decision Support pattern creates four voices around one question: a core Claude voice plus Skeptic, Pragmatist, and Critic perspectives. That mix helps surface hidden assumptions, failure modes, and operational tradeoffs fast. The value is not more content; it is better friction.

Who should install it

Install council if you regularly make ambiguous decisions in product, engineering, operations, or strategy and need a repeatable way to pressure-test choices. It is especially useful when stakeholders disagree, the cost of a wrong call is non-trivial, or you want a more defensible recommendation than a generic “pros and cons” list.

When it is a strong fit

Use council when the question has at least two credible paths, the decision depends on context, or you need a go/no-go call with dissent built in. It is less useful for straightforward execution, bug fixing, architecture planning, or tasks where the answer is already clear and you just need action.

How to Use council skill

Install the council skill

Add council to your Claude Code skills set with the repository’s install flow, then invoke it when you need structured decision support rather than a standard completion. For this repo, the practical install reference is the skills/council path in affaan-m/everything-claude-code. After installation, the skill should be available as council in the same environment where you work on prompts and tasks.

Shape a prompt the skill can debate

The best council usage starts with a decision statement, not a vague topic. Give the choice, the stakes, and the constraint that matters most. Strong input looks like this: “Should we ship feature flags now or wait for a fuller rollout plan? Priority is minimizing operational risk while preserving launch date.” Weak input looks like: “Thoughts on feature flags?” The first lets council evaluate tradeoffs; the second invites generic advice.

Read the right files first

Start with SKILL.md to understand the decision boundary, then inspect the sections that define when to use council, when not to use it, the role structure, and the workflow. Those are the parts that most affect whether council is the right tool. If you are adapting it into your own repository, read the surrounding skill conventions so the output style matches your environment.

Use a workflow, not a one-off question

A good council workflow is: define the decision, list 2-4 candidate options, state constraints, run the council, then ask for a recommendation with rationale and risks. Include operational realities such as deadlines, user impact, reversibility, and what would make the decision fail. That gives the Pragmatist and Critic enough context to add real value.

council skill FAQ

Is council for brainstorming or deciding?

It is for deciding. The council skill is designed to resolve ambiguity by making disagreement explicit, not to generate an endless list of ideas. If you only need raw ideation, a simpler prompt is usually enough. Use council when you need a recommendation after the options have been pressure-tested.

How is council different from a normal prompt?

A normal prompt often collapses multiple viewpoints into one answer. council forces contrasting lenses to speak separately, which is useful when the risk is premature consensus. In practice, that means better handling of tradeoffs, fewer hidden assumptions, and a clearer sense of what could go wrong.

Is council good for beginners?

Yes, if the user can name the decision clearly. Beginners get the most value when they provide a concrete choice, a short context summary, and the constraint they care about most. If the prompt is fuzzy, the skill can still work, but the debate will be less decisive.

When should I not use council?

Do not use council for simple factual questions, obvious tasks, code review, security review, or step-by-step implementation planning. The repository explicitly points those cases to other skills or to direct execution. council for Decision Support is strongest only when judgment, not mechanics, is the hard part.

How to Improve council skill

Give the council better decision inputs

The highest-leverage improvement is sharper context. State the options, the decision deadline, the main risk, and the success criterion. For example: “Choose between A and B; A is faster, B is safer; deadline is Friday; success means no rollback and minimal support load.” This is much stronger than describing the problem in abstract terms.

Ask for dissent on the real constraint

If you care about cost, reversibility, user experience, or operational load, say so up front. council works best when it knows what to attack. Without that, the Skeptic and Critic may challenge the wrong assumptions, and the recommendation will be less actionable.

Iterate after the first run

If the first council answer is too broad, rerun it with a narrower question or a different constraint, such as “optimize for speed,” “optimize for safety,” or “optimize for long-term maintainability.” That is often more useful than asking the same prompt again. For council skill, better iteration means better framing, not just more words.