agentleFS
Sign inStart free
juandelossantosjuandelossantos/another-agent-skills/skills/idea-refine/SKILL.md

idea-refine

Skill1 starsChanged 11 days agoMITView on GitHub

Refine raw ideas into sharp, actionable concepts through divergent and convergent thinking. Use when an idea is vague or needs stress-testing. Do NOT use when requirements are already clear.

We split this skill into the rules it gives an agent

1 rule. 0 are common, 0 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 53 edited versions we read keep these.

  • 0

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use techniques from guides/DIVERGENT-CONVERGENT.md:

    line 59

“Common” is measured: kept by at least 6 in 10 of the 53 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

Idea Refine

Turn vague concepts into actionable specifications.

Sits between interview-me (extracting intent) and spec-driven-development (writing specs). Uses structured divergent thinking (expand options) followed by convergent thinking (select best path).

Pipeline Position

interview-me → idea-refine (this skill) → spec-driven-development

When to Use

  • Idea is still vague or incomplete
  • Need to stress-test assumptions before committing
  • Want to explore alternatives before converging on one

When NOT to Use

  • Requirements are already clear and documented
  • Trivial changes with obvious scope

Output Contract

Refined concept + selected direction — documented alternatives (3+) from divergent phase, selection criteria from convergent phase, assumption audit with must-be-true statements, clear rationale for chosen direction, ready to feed into spec-driven-development.

Core Process: Double Diamond

Adapted from Design Council's framework. Two diamonds, each with diverge + converge:

Diamond 1 — Problem Space:
  Diverge: Explore alternatives, challenge assumptions, expand options
  Converge: Define the specific problem, select direction

Diamond 2 — Solution Space:
  Diverge: Generate solution variants, combine approaches
  Converge: Select best solution, prepare for spec

Phase 1 — Diverge (Expand Options)

Goal: generate span, not depth. Resist evaluating during this phase.

Use techniques from guides/DIVERGENT-CONVERGENT.md:

TechniqueBest For
SCAMPERReimagining existing approaches
Reverse ThinkingBreaking out of incremental ruts
First PrinciplesDecomposing to fundamentals
How Might WeReframing the problem
Worst IdeaLowering inhibition, unlocking creativity

Facilitation prompts:

  • "What if we did the opposite?"
  • "What if budget/time/tech weren't factors?"
  • "Who else has this problem? How do they solve it?"
  • "What would this look like at 10x scale?"
  • "What if it could only have one feature?"

Phase 2 — Converge (Select Path)

Goal: filter, prioritize, commit. Use evidence, not instinct.

TechniqueBest For
Impact/Effort MatrixPrioritizing by value vs cost
NUF TestScoring novelty, usefulness, feasibility
Assumption AuditSurfacing unvalidated bets
Pre-mortemStress-testing chosen direction

Facilitation prompts:

  • "What could kill this idea? If it failed in 12 months, why?"
  • "What must be true for this to work?"
  • "Would users switch from their current approach?"
  • "Which failure modes are acceptable? Which are dealbreakers?"

Output

The selected direction feeds directly into spec-driven-development as the starting point for SPEC.md.

Anti-Patterns

  1. Premature convergence — Picking the first idea without alternatives explored.
  2. Anchoring — First idea (especially from senior person) becomes default.
  3. Confirmation bias — Seeking evidence that supports preferred idea, dismissing counter-evidence.
  4. Solution-anchored framing — "How do we add X?" instead of "What problem are we solving?"
  5. Skipping assumption audit — Moving to spec without surfacing what must be true.
  6. Yes-machine facilitation — Never pushing back on weak ideas.

Verification

  • Divergent phase completed (2+ techniques used, 3+ alternatives generated)
  • Convergent phase completed (selection criteria defined and applied)
  • Assumption audit surfaced must-be-true statements
  • Selected direction has clear rationale (not just "felt right")
  • Output ready to feed into spec-driven-development
  • bash scripts/skill-lint.sh skills/idea-refine passes

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