| name | woos-idea-capture |
| description | Capture and structure raw ideas through guided interview or quick note. Produces a structured idea document ready for research or PRD pass. Focuses purely on product intent โ no technical decisions. |
| version | 3.0.0 |
| author | Hermes Profile |
| license | MIT |
| metadata | {"hermes":{"tags":["idea","capture","interview","product","ideation"],"related_skills":["woos-product-discovery","woos-product-design-flow"]}} |
Idea Capture
Output Language
- Author all user-facing prose (narratives, findings, summaries) in the user's most recent input language (e.g. if the user is writing in Chinese, write the PRD body, review findings, and summaries in Chinese).
- Keep these tokens verbatim in English regardless of the user's language, because they are contract identifiers consumed by other skills and gates:
- Verdict tokens:
PASS, REQUEST_CHANGES, BLOCKED, NOT_RUN
- Severity tokens:
critical, high, medium, low, warning
- Tag tokens:
[ASSUMPTION: ...], [NEEDS CLARIFICATION: ...], [NEEDS INPUT: ...]
- Phase / dimension IDs:
P0 โฆ P11, P2a, Phase A, Phase B
- Shape values:
internal-tool, single-operator, consumer-product, multi-stakeholder, CLI
- P0 rating values:
strong, adequate, thin, broken
- Section headings declared in templates (e.g.
## Background, ## Functional Requirements, ## Assumptions Index) โ translations break downstream structural checks
- Field labels inside templates (e.g.
**Consequences (testable):**, **Out of Scope:**, **User value:**, Given โฆ When โฆ Then โฆ)
- Code, file paths, IDs, and quoted PRD excerpts stay in their original form.
Purpose
Transform a raw idea or feature request into a structured document that can feed into research, PRD, or product planning.
This skill focuses exclusively on product intent: what problem to solve, for whom, and what success looks like. It does NOT make technical decisions โ those belong in later stages.
When to Use
Use when:
- User presents a new idea, feature request, or product vision
- Starting the woos-product-discovery flow at Step 1
- Need to clarify scope before committing to a full PRD
Skip when:
- Idea already has a PRD โ go to
woos-product-design-flow
- Pure bugfix with clear reproduction โ no capture needed
Capture Process
Note: Capture itself always runs the same way, but mode selection happens in two places:
- Lite may branch immediately after Capture if the idea is obviously trivial and the user confirms.
- Standard / Strict are inferred later, after Product Discovery, from the approved roadmap.
Two capture depths based on idea complexity:
Quick Note (trivially simple ideas)
For: single-line fixes, obvious improvements, user already has full clarity.
Process:
- Record a concise idea summary
- Identify target user, core problem, and proposed solution
- Note constraints if any
- Output to
ideas/<slug>.md
Quick Note template:
# <Idea Title>
## Summary
One paragraph: what and why.
## Problem
Who has this problem? What pain does it cause?
## Proposed Solution
What should we build?
## Constraints (if any)
- Time, budget, technical, or compatibility constraints
## Success Signal
How do we know it worked?
Guided Interview (anything non-trivial)
For: features, product initiatives, anything requiring design thinking.
Process:
- Run through interview questions one at a time
- Capture answers in structured format
- User can say
GREENLIGHT NEXT STAGE to skip remaining questions
- Output to
ideas/<slug>/00-idea-capture.md
Interview Question Bank
Ask sequentially. Skip if user already provided the information unprompted.
Category 1: Problem & Users
- Who will use this? (Target persona / user type)
- What problem does it solve? (Pain point, frequency, severity)
- How do they solve it today? (Current workaround, if any)
Category 2: Vision & Behavior
- What should the ideal experience look like? (User's mental model)
- What are the core behaviors? (Must-have interactions / flows)
- What would make this feel "done" to you? (Definition of success)
Category 3: Scope & Boundaries
- What's explicitly out of scope? (Things you do NOT want built)
- Are there time or resource constraints? (Deadline, budget, team size)
- Does this need to work on day one, or can it be iterative? (MVP vs complete)
Category 4: Context & Dependencies
- Does this need to connect to anything that already exists? (Existing products, data sources, user accounts โ NOT tech stack)
- Any security or compliance concerns? (Data sensitivity, access control, regulatory)
Category 5: Quality & Risk
- What would make this fail? (Risks, failure modes, deal-breakers)
- How will you know this is successful? (Observable outcomes, metrics)
Handling Technical Preferences
Users often volunteer technical opinions during capture ("I want to use Go", "let's use SQLite", "no React"). This is natural โ but it must NOT become a hard constraint in this phase.
Rules:
- Do NOT ask questions about tech stack, frameworks, languages, or databases
- If user volunteers a tech preference โ acknowledge it, then record it in the output under a dedicated
## Technical Preferences (Deferred) section
- Label clearly โ these are preferences, not decisions. They will be evaluated during the Discovery architecture steps (Step 4 / Step 4R)
- Do NOT let preferences leak into Problem Statement, Vision, Core Behaviors, or Constraints sections
- Do NOT push back on preferences โ just record and defer. No debate in this phase.
Output format for deferred preferences:
## Technical Preferences (Deferred)
> โ ๏ธ These are user preferences expressed during idea capture. They are NOT confirmed
> technical decisions. They will be evaluated during the Discovery architecture
> steps (Step 4 / Step 4R), where trade-offs can be properly assessed.
- "Backend in Go" โ user preference, to be evaluated in architecture
- "SQLite for storage" โ user preference, to be evaluated in architecture
Output Format (Standard / Strict Mode)
# <Idea Title>
## Capture Date
YYYY-MM-DD
## Source
How did this idea come up?
## Problem Statement
Who: [target users]
Pain: [what hurts]
Current workaround: [how they cope today]
## Vision
Ideal experience: [describe]
## Core Behaviors
1. [behavior 1]
2. [behavior 2]
## Success Criteria
- [observable outcome 1]
- [observable outcome 2]
## Out of Scope
- [explicitly excluded]
## Constraints
- Time: [deadline or "none"]
- Resources: [team size or "flexible"]
- Dependencies: [integration points or "none"]
- Security: [concerns or "standard"]
## Technical Preferences (Deferred)
> โ ๏ธ User preferences only. NOT decisions. Evaluated in Discovery Step 4 / Step 4R (Architecture).
- [preference]: user preference, to be evaluated in architecture
- (or "None expressed")
## Risks
- [risk]: [mitigation or "accepted"]
## Open Questions
- [unresolved question]
## Interview Status
- Questions answered: N/13
- GREENLIGHT used: yes/no
File Locations
- Quick Note (trivial):
ideas/<slug>.md
- Guided Interview:
ideas/<slug>/00-idea-capture.md
Slug: lowercase, hyphens, max 50 chars, from idea title.
After Capture โ Lite Branch Decision
After capture is complete, assess complexity:
Is it obviously trivial? (typo, 1-liner, single obvious change, user explicitly says it's simple)
โ Yes โ Propose Lite to user: "This looks straightforward. I suggest Lite mode โ go directly to a lightweight PRD package. Agree?"
โ User confirms โ `woos-product-design-flow` Lite
โ User says no โ proceed to Discovery
โ No โ proceed to `woos-product-discovery` (research needed)
Rules:
- Lite is a PROPOSAL, not a default. User must confirm.
- If there are open questions, risks, or dependencies โ NOT trivial โ Discovery.
- When in doubt, proceed to Discovery. Lite is only for the obviously simple.
Pitfalls
- Don't re-ask questions the user already answered unprompted
- Don't prescribe architecture โ capture requirements, not solutions
- Don't skip open questions โ they feed the research pass
- Don't write the PRD here โ that's handled in
woos-product-design-flow
- Don't make technical decisions โ those belong in later stages
- Don't discuss tech stack, frameworks, or databases โ stay on product intent
- Don't auto-select Lite without user confirmation
- Don't ASK about tech stack โ if user volunteers preferences, record under "Technical Preferences (Deferred)", never in Constraints
- Don't let tech preferences become constraints โ "user wants Go" โ "must use Go". Record as preference, let architect evaluate in Discovery Step 4 / Step 4R
- Don't debate tech choices โ this phase has no authority to confirm or reject technical decisions