SuperAPI AI Directory

problem-framing

General & other

View repository

Updated 10-16 05:45

About this skill

Problem Framing

Turn an approved intake route into a crisp problem statement and a small set of decision-ready options.

Context

Problem framing sits immediately after intake when routing is clear but the work is still underspecified. It absorbs the strongest part of brainstorming-style collaboration without collapsing the entry gate into a full design workshop.

Use it to answer:

  • What problem are we actually solving?
  • What boundaries must stay fixed?
  • What are the viable directions?
  • Which direction should the team carry into specification, research, or architecture?

Do not use it when the route is already well-scoped enough for downstream work, or when the user is already in implementation, debugging, or review.

Inputs

  • intake-brief -- Must identify the work type, entry phase, recommended workflow, key risks, quality_target_context, and the next likely skill. Preserve the intake language boundary fields instead of silently guessing them away: source_language, artifact_record_language, and user_presentation_locale. Preserve quality_target_context so downstream quality and security work does not reconstruct the runtime or exposure boundary from guesses.

Process

Step 1: Read the Intake Route

Review the intake brief before asking anything new.

Extract:

  • the requested outcome
  • the routing rationale
  • the risks and unknowns already identified
  • the downstream phase that this framing is meant to unlock

Step 2: Ask Only Decision-Changing Questions

Ask questions one at a time.

Default budget:

  • 1-3 questions in normal cases
  • up to 5 only if each additional answer materially changes scope, risk, or option selection

Focus on:

  • success criteria
  • hard constraints
  • non-goals
  • irreversible decisions or dependencies

Do not restate intake questions unless the answer is still ambiguous or conflicting.

Step 3: Write the Problem Frame

Produce a concise problem-frame covering:

  • problem statement
  • target users or operators
  • constraints
  • non-goals
  • assumptions
  • open questions

The canonical problem-frame record stays in English under current repo policy, but user-facing framing output may still be presented in the user's locale. Carry source_language, artifact_record_language, and user_presentation_locale forward explicitly.

Carry quality_target_context forward explicitly when it affects scope, risk, or downstream QA. If the target is an agent-internal skill, host runtime tool, or local harness, do not let problem framing drift into a public service design unless the user chose that product target.

The problem frame must be sharp enough that downstream skills do not have to rediscover the core problem. Use plain language, present user-facing output in the user's requested language or the user_presentation_locale, and explicitly note system shape or collaboration quality when they materially affect the framing.

Step 4: Compare 2-3 Directions

Produce an options-brief with 2-3 plausible directions.

For each direction, note:

  • why it is viable
  • what it optimizes for
  • what it risks or defers
  • when it should be rejected

Keep these at the level of product or solution direction, not low-level implementation detail.

Step 5: Recommend One Direction

Record a design-direction that includes:

  • the recommended option
  • why it wins over the alternatives
  • what must remain open for downstream skills
  • the next lifecycle destination (market-analysis, user-research, feasibility-study, or requirements-engineering)

When the route depends on team boundaries, ownership clarity, workflow friction, or architectural sprawl, say so directly as a collaboration quality or system shape concern instead of hiding it inside generic risk wording.

Step 6: Get Approval and Handoff

Ask the user to confirm the framing and chosen direction.

After approval, hand off the framing artifacts to the next skill named in the design-direction.

Outputs

  • problem-frame -- The clarified problem, constraints, and non-goals
  • options-brief -- Small set of viable directions with trade-offs
  • design-direction -- Approved recommendation plus next lifecycle destination

Quality Gate

  • problem-frame states the problem, constraints, non-goals, assumptions, and open questions
  • options-brief compares 2-3 viable directions with explicit trade-offs
  • design-direction recommends one direction and names the next skill to invoke
  • The total question load stayed within the default budget or justified why it exceeded it
  • The user approved the framing output before handoff

Anti-Patterns

  1. Repeating intake -- Intake already handled classification and routing. Framing should deepen the decision, not restart the entry gate.
  2. Premature architecture -- Stay at the level of direction and trade-off. Leave system structure to system-design.
  3. Infinite discovery interviews -- Use the minimum questions needed to make a decision. If more information is required, route to research skills.
  4. Single-option framing -- If only one path is shown, there is no real trade-off analysis.

Related Skills

  • intake -- provides the approved route and initial constraints
  • market-analysis -- use when market uncertainty remains high
  • user-research -- use when user behavior or pain points are still unclear
  • feasibility-study -- use when go/no-go needs deeper validation
  • requirements-engineering -- consumes the approved design direction

Distribution

  • Public install surface: skills/.curated
  • Canonical authoring source: skills/00-discovery/problem-framing/SKILL.md
  • This package is exported for npx skills add/update compatibility.
  • Packaging stability: beta
  • Capability readiness: core