Skip to main content

Define the clinical need

Install this skill
npx skills add StanfordSpezi/SpeziVibe --skill biodesign-needs-finding

Or install all skills: npx skills add StanfordSpezi/SpeziVibe --skill '*'. See the Getting Started guide for tool-specific options.

Explore this skill in browser chat. Each button opens a new conversation; bring any earlier briefs with you and save the results yourself.

Need to paste the prompt yourself?
Browser prompt for biodesign-needs-finding
Please act as the SpeziVibe `biodesign-needs-finding` skill and walk me through it interactively.

STEP 1 — Fetch the skill instructions from this URL and read them carefully:
https://raw.githubusercontent.com/StanfordSpezi/SpeziVibe/main/skills/biodesign-needs-finding/SKILL.md

If the skill references other files or other skills by name, fetch them from the same repository:
- Files inside this skill (e.g. `references/foo.md`): https://raw.githubusercontent.com/StanfordSpezi/SpeziVibe/main/skills/biodesign-needs-finding/<relative-path>
- Other skills (e.g. `biodesign-needs-finding`): https://raw.githubusercontent.com/StanfordSpezi/SpeziVibe/main/skills/<skill-name>/SKILL.md

STEP 3 — Walk me through the skill interactively. When a step would normally save a markdown file to a project, show the file content in a code block instead so I can copy it. Don't simulate my answers — wait for me to respond.

If you cannot fetch the skill instructions or their reference files, ask me to paste them before proceeding.
Ask me to paste or attach any existing briefs you need. Do not claim to have read files, run commands, created a project, or tested an integration from this chat. Help me prepare the decisions and instructions, then hand off file and code work to my coding tool.

If your chat can’t open the skill link, open the instructions in a new tab and paste them into the chat.

Walks you through a Stanford Biodesign-style needs-finding process to define a clear, solution-free problem statement.

How It Works​

This is a Socratic coaching process — it asks questions and challenges your thinking rather than writing the statement for you. It validates each component: the health problem, the affected population, and the desired outcome.

If you start with a technology-first statement like "We need an app for...", it redirects you to the underlying problem.

Output​

Produces docs/planning/need-statement.md containing:

  • A structured need statement: "A way to [PROBLEM] in [POPULATION] in order to [OUTCOME]"
  • Supporting evidence and validation notes
  • Key stakeholders to validate with
  • Identified insights about why the need is unmet

Example​

AI:   Do you already have a need statement or draft, or are you
starting fresh?

You: We need an app for cardiac rehab patients

AI: That sounds like a solution — let's work backward to the
underlying problem. What's happening with cardiac rehab
patients that isn't working today?

You: They don't stick with their exercise programs after discharge

AI: Good — that's the problem space. Which patients specifically?
Post-MI? Post-CABG? Heart failure? And at what point after
discharge does adherence drop off?

...

AI: Here's your refined need statement:
"A way to sustain exercise adherence in post-discharge
cardiac rehabilitation patients in order to reduce
30-day readmission rates and improve functional recovery."

Limitations​

  • Does not write the statement for you — requires your active participation
  • If you skip quality checks, tracks them as "skipped-step risks" in the output