agentleFS
theultimate-devtheultimate-dev/skills/skills/product-design/guiding-product-discovery/SKILL.md

guiding-product-discovery

Skill2 starsChanged 47 days agoMITView on GitHub

Guide a rough product, application, feature, or proof-of-concept idea through requirements, UX flows, visual directions, offline HTML prototypes, and a product/design handoff. Use to coordinate discovery across phases or resume unfinished design work; individual artifact requests belong to the relevant specialist.

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 4 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    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. Distinctive:

    Read applicable project instructions, including AGENTS.md or CLAUDE.md when present, and inspect existing briefs, source, design assets, …

    line 13
  2. Very specific:

    Read workflow and milestones to choose the next useful phase.

    line 19
  3. Distinctive:

    Use the host's supported skill invocation mechanism, or read an available skill and carry out its instructions.

    line 31
  4. Distinctive:

    If a specialist is unavailable, name the missing capability and provide the portable task brief, including enough context to continue …

    line 33
  5. Distinctive:

    Keep the index small: current artifacts, approved choices, assumptions, unresolved questions, and next action.

    line 45

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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

Guiding product discovery

Help the user make consequential product and design choices while agents develop concrete proposals between decisions. Finish with a product/design handoff, not production implementation or a silently selected architecture.

Start from the project

Read applicable project instructions, including AGENTS.md or CLAUDE.md when present, and inspect existing briefs, source, design assets, tokens, and decisions. In a chat without repository access, use supplied materials and identify what could not be inspected. Do not ask the user to repeat discoverable information.

Capture the intended audience, problem, desired outcome, boundaries, and unresolved questions. An established stack constrains feasibility; a missing stack does not block discovery. Raise a stack decision only when it changes a proposal's feasibility or the handoff. Never replace project technology by preference.

For an existing product, establish whether this initiative preserves, evolves, or replaces its visual language. Infer this from explicit instructions already given; otherwise ask with concrete implications. Carry the choice forward, including any approved exceptions.

Read workflow and milestones to choose the next useful phase. Adopt existing artifact locations. Otherwise use docs/product-design/<initiative>/ in the target project and the discovery index. The path is an output convention, not an installation requirement.

Coordinate by outcome

Needed outcomeSpecialist, when availableRequired input
Agreed problem, scope, requirementsshaping-product-briefsIdea, known audience, constraints, evidence
Journeys, information hierarchy, screen behaviordesigning-ux-flowsBrief or equivalent requirements
Three comparable visual direction briefsexploring-visual-directionsAudience, content, flows, visual continuity choice
Three offline sketches, then a refined selected prototypebuilding-html-prototypesDirection briefs or selected direction and flows
Product/design handoff and suggested slicespreparing-implementation-handoffsCurrent requirements, flows, chosen design, prototype and review findings

Use the host's supported skill invocation mechanism, or read an available skill and carry out its instructions. This workflow does not require subagents, concurrent execution, or a particular tool. Do not invent skill paths or assume that installing this skill installed the specialists.

If a specialist is unavailable, name the missing capability and provide the portable task brief, including enough context to continue elsewhere. Continue independent discovery work; do not label an unperformed specialist step complete.

Keep the user in the decisions

Default to three milestones: agree product scope; select or combine a direction after reviewing three HTML sketches; review the refined prototype and readiness for handoff. Prepare a recommendation and concrete alternatives before asking. Existing approval counts: do not ask again unless new information materially changes the choice. A request for only one artifact does not authorize the whole workflow.

Between milestones, resolve routine details autonomously within the agreed scope. Ask when an unresolved choice materially changes outcomes, effort, or the experience. Record which decisions are the user's, which are provisional recommendations, and what remains unknown. Never treat silence or a finished-looking prototype as approval.

Three sketches are the default for unresolved visual direction, not a ritual for every change. Skip exploration already settled by the user. Merge documents for small work and omit irrelevant states or diagrams. Read the worked examples when calibrating effort for a new application, small feature, or proof of concept.

Resume and revise

Keep the index small: current artifacts, approved choices, assumptions, unresolved questions, and next action. Give each requirement a stable local identifier when needed for traceability. Other artifacts reference the requirement instead of copying its text everywhere.

When a prototype reveals a problem, revise affected requirements or flows first, then update the direction, prototype, and handoff as needed. Mark a prior approval as needing review only where the change invalidates it. Retain rejected directions and the reasons they were rejected when useful for future decisions.

Finish

Return links to current artifacts, the key decisions, any remaining blockers, and the next engineering questions. Product/design readiness means sufficient agreed behavior and visual intent to begin engineering planning; it does not assert architectural readiness, production accessibility conformance, or research validation.

Without file-writing capability, provide complete copyable artifacts and filenames and state that no files were written. Never claim visual or interaction verification unless it was performed.

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.