/skill:ideate — Explore and Shape Ideas
Purpose: Take a rough idea, a single document, or a folder of related documents and turn them into a clear direction with identified options, risks, and next steps. Does NOT produce implementation plans — use superpowers:brainstorming or superpowers:writing-plans after the direction is chosen.
When to Use
- Rough idea or scattered docs need shaping into a coherent direction
- Deciding between multiple technical or business approaches
- Upstream of
superpowers:writing-plans or superpowers:brainstorming, when the problem itself isn't framed yet
When NOT to Use
- Direction is already clear — jump to
superpowers:brainstorming or superpowers:writing-plans
- You need an implementation plan — use
superpowers:writing-plans
- Requirements exist but are ambiguous — use
/skill:prd-validate
Arguments
<mode> — optional first arg: idea (default), ba, tech
<path-or-description> — document path, folder path, or plain-language idea
Examples:
/skill:ideate "We need a better enterprise onboarding experience"
/skill:ideate ba docs/discovery/
/skill:ideate tech docs/architecture/integration-notes.md
Modes
| Mode |
When to use |
Deepens into |
idea (default) |
General idea shaping and option framing |
Concept, options, what to validate next |
ba |
Business process, stakeholder alignment, workshop prep |
Process, scope, assumptions, dependencies, workshop outputs |
tech |
Technical directions, architecture options, systems impact |
Constraints, integration, architecture tradeoffs, validation steps |
If no mode is supplied, default to idea. Lean toward ba when the ask is primarily about business process or stakeholder alignment.
Input Modes
Single document
When the user provides one artifact (BRD, PRD, story, notes, proposal, architecture doc):
- Extract what the document already says
- Identify what it implies
- Surface gaps and open questions
- Turn it into a clearer ideation output
Folder of documents
When the user provides a folder:
- When the
jev-gate MCP server is available (see _internal/jev-gate), call triage_coding_task(task_prompt, candidate_files) to score and prune candidate files (keep_full, keep_signatures_only, drop) before loading documents into context
- Identify the most relevant files (don't read everything)
- Synthesize repeated themes
- Surface contradictions and missing decisions
- Name which files drove the conclusions
- When external technical, hardware, CAD/slicer, or SDK lookups are needed during
tech or idea shaping, call smart_search(query, force_backend="auto") via jev-gate before executing any web search (escalating to Perplexity only when needs_perplexity_synthesis >= 0.80)
Blank start
When the user has only a rough idea:
"Tell me what you want to achieve in plain language. I'll help shape it into something we can explore with stakeholders."
Shared Starting Flow (All Modes)
Always establish the lightest useful version of:
- Problem or opportunity
- Affected users or stakeholders
- Desired outcome
- Known facts
- Assumptions and risks
- Major open questions
- Plausible directions
Then branch by mode.
Interaction Rules
- Ask one question at a time
- Prefer multiple-choice with a recommended default when practical
- Start with high-leverage ambiguity, not formatting
- Distinguish facts, assumptions, risks, and open questions
- If the user says "you choose," propose a default and explain the tradeoff
- Do not force implementation details during early ideation
Outputs (lightest useful set)
Idea framing
- Problem statement, affected users, desired outcome, success signals
Scope shaping
- In scope, out of scope, MVP, later-phase ideas
Option analysis (when multiple directions are plausible)
- Summary, benefits, tradeoffs, delivery/data/integration impact, recommendation
Assumptions and gaps
- Known facts, assumptions, risks, dependencies, open decisions
BA mode additions
- Current/future state, process friction, operational changes
- Workshop-ready: problem framing, stakeholder question set, decision agenda, parking lot
Tech mode additions
- Systems touched, ownership boundaries, affected interfaces
- Data flows, source-of-truth concerns, integration patterns
- Major technical risks, spikes/experiments worth running
- Technical workshop brief: objective, systems in play, decision areas
Output Patterns
Quick synthesis
Idea summary
What appears to be true
- ...
What is still assumed
- ...
What needs a decision
- ...
Recommended next step
- ...
Workshop brief
- Objective, participants, context summary, key questions, option areas, decisions to leave with
Direction recommendation
- Recommendation, why this direction, tradeoffs to accept, what to validate next
Handoff Guidance
At the end, recommend the next workflow:
- Continue ideation (more questions to resolve)
- Write a PRD →
/skill:prd-create; upgrade an existing one → /skill:prd-validate
- Design the solution →
superpowers:brainstorming
- Plan the implementation →
superpowers:writing-plans
- Do not jump to
superpowers:writing-plans until the direction is clear enough to plan
CRITICAL: Do Not Guess
- Do NOT invent certainty not supported by the source material or user input.
- Do NOT read every file in a folder if a smaller set establishes the pattern.
- Do NOT force technical implementation details during early ideation unless they materially change the business decision.
- Do NOT turn routine implementation defaults into workshop decisions.
- Do NOT jump to
superpowers:writing-plans until the direction is chosen.
- Do NOT lose shared ideation context when switching from
idea into ba or tech mode.
Contract
- Inputs: a rough ask, a single document path, or a folder of related documents; mode flag (
idea | ba | tech, default idea).
- Preconditions: human operator at the keyboard (this is interactive ideation, not autonomous synthesis); inputs are readable.
- Outputs: a direction document with identified options, risks, closed decisions where settled, and explicit open questions. Hands off to
/skill:prd-create (write a new PRD), /skill:prd-validate (upgrade an existing one), superpowers:writing-plans, or superpowers:brainstorming once the direction is chosen.
- Postconditions: the artifact is concrete enough to drive
/skill:prd-validate or superpowers:writing-plans; routine implementation defaults are NOT promoted into workshop decisions; mode-switch transitions preserve shared context.
- Failure modes: input too vague for any mode → ask one clarifying question and stop; multi-document inputs disagree on direction → surface the conflict, do not silently average them.
1---2name: ideate-23description: 'Use before superpowers:writing-plans or brainstorming to shape a rough4 ask, doc, or folder into a direction. Three modes: idea (general), ba (business-analysis),5 tech (technical options).'6whenToUse: 'Use before superpowers:writing-plans or brainstorming to shape a rough7 ask, doc, or folder into a direction. Three modes: idea (general), ba (business-analysis),8 tech (technical options).'9metadata:10 source: https://skillmd.com/skills/garyld1962/ideate-211---121314# /skill:ideate — Explore and Shape Ideas1516**Purpose:** Take a rough idea, a single document, or a folder of related documents and turn them into a clear direction with identified options, risks, and next steps. Does NOT produce implementation plans — use `superpowers:brainstorming` or `superpowers:writing-plans` after the direction is chosen.1718## When to Use1920- Rough idea or scattered docs need shaping into a coherent direction21- Deciding between multiple technical or business approaches22- Upstream of `superpowers:writing-plans` or `superpowers:brainstorming`, when the problem itself isn't framed yet2324## When NOT to Use2526- Direction is already clear — jump to `superpowers:brainstorming` or `superpowers:writing-plans`27- You need an implementation plan — use `superpowers:writing-plans`28- Requirements exist but are ambiguous — use `/skill:prd-validate`2930## Arguments3132- `<mode>` — optional first arg: `idea` (default), `ba`, `tech`33- `<path-or-description>` — document path, folder path, or plain-language idea3435Examples:36```37/skill:ideate "We need a better enterprise onboarding experience"38/skill:ideate ba docs/discovery/39/skill:ideate tech docs/architecture/integration-notes.md40```4142## Modes4344| Mode | When to use | Deepens into |45|---|---|---|46| `idea` (default) | General idea shaping and option framing | Concept, options, what to validate next |47| `ba` | Business process, stakeholder alignment, workshop prep | Process, scope, assumptions, dependencies, workshop outputs |48| `tech` | Technical directions, architecture options, systems impact | Constraints, integration, architecture tradeoffs, validation steps |4950If no mode is supplied, default to `idea`. Lean toward `ba` when the ask is primarily about business process or stakeholder alignment.5152## Input Modes5354### Single document5556When the user provides one artifact (BRD, PRD, story, notes, proposal, architecture doc):57- Extract what the document already says58- Identify what it implies59- Surface gaps and open questions60- Turn it into a clearer ideation output6162### Folder of documents6364When the user provides a folder:65- When the `jev-gate` MCP server is available (see `_internal/jev-gate`), call `triage_coding_task(task_prompt, candidate_files)` to score and prune candidate files (`keep_full`, `keep_signatures_only`, `drop`) before loading documents into context66- Identify the most relevant files (don't read everything)67- Synthesize repeated themes68- Surface contradictions and missing decisions69- Name which files drove the conclusions70- When external technical, hardware, CAD/slicer, or SDK lookups are needed during `tech` or `idea` shaping, call `smart_search(query, force_backend="auto")` via `jev-gate` before executing any web search (escalating to Perplexity only when `needs_perplexity_synthesis >= 0.80`)7172### Blank start7374When the user has only a rough idea:75> "Tell me what you want to achieve in plain language. I'll help shape it into something we can explore with stakeholders."7677## Shared Starting Flow (All Modes)7879Always establish the lightest useful version of:80- Problem or opportunity81- Affected users or stakeholders82- Desired outcome83- Known facts84- Assumptions and risks85- Major open questions86- Plausible directions8788Then branch by mode.8990## Interaction Rules9192- Ask one question at a time93- Prefer multiple-choice with a recommended default when practical94- Start with high-leverage ambiguity, not formatting95- Distinguish facts, assumptions, risks, and open questions96- If the user says "you choose," propose a default and explain the tradeoff97- Do not force implementation details during early ideation9899## Outputs (lightest useful set)100101### Idea framing102- Problem statement, affected users, desired outcome, success signals103104### Scope shaping105- In scope, out of scope, MVP, later-phase ideas106107### Option analysis (when multiple directions are plausible)108- Summary, benefits, tradeoffs, delivery/data/integration impact, recommendation109110### Assumptions and gaps111- Known facts, assumptions, risks, dependencies, open decisions112113### BA mode additions114- Current/future state, process friction, operational changes115- Workshop-ready: problem framing, stakeholder question set, decision agenda, parking lot116117### Tech mode additions118- Systems touched, ownership boundaries, affected interfaces119- Data flows, source-of-truth concerns, integration patterns120- Major technical risks, spikes/experiments worth running121- Technical workshop brief: objective, systems in play, decision areas122123## Output Patterns124125### Quick synthesis126```127Idea summary128129What appears to be true130- ...131132What is still assumed133- ...134135What needs a decision136- ...137138Recommended next step139- ...140```141142### Workshop brief143- Objective, participants, context summary, key questions, option areas, decisions to leave with144145### Direction recommendation146- Recommendation, why this direction, tradeoffs to accept, what to validate next147148## Handoff Guidance149150At the end, recommend the next workflow:151- Continue ideation (more questions to resolve)152- Write a PRD → `/skill:prd-create`; upgrade an existing one → `/skill:prd-validate`153- Design the solution → `superpowers:brainstorming`154- Plan the implementation → `superpowers:writing-plans`155- Do not jump to `superpowers:writing-plans` until the direction is clear enough to plan156157## CRITICAL: Do Not Guess158159- Do NOT invent certainty not supported by the source material or user input.160- Do NOT read every file in a folder if a smaller set establishes the pattern.161- Do NOT force technical implementation details during early ideation unless they materially change the business decision.162- Do NOT turn routine implementation defaults into workshop decisions.163- Do NOT jump to `superpowers:writing-plans` until the direction is chosen.164- Do NOT lose shared ideation context when switching from `idea` into `ba` or `tech` mode.165166## Contract167168- **Inputs:** a rough ask, a single document path, or a folder of related documents; mode flag (`idea` | `ba` | `tech`, default `idea`).169- **Preconditions:** human operator at the keyboard (this is interactive ideation, not autonomous synthesis); inputs are readable.170- **Outputs:** a direction document with identified options, risks, closed decisions where settled, and explicit open questions. Hands off to `/skill:prd-create` (write a new PRD), `/skill:prd-validate` (upgrade an existing one), `superpowers:writing-plans`, or `superpowers:brainstorming` once the direction is chosen.171- **Postconditions:** the artifact is concrete enough to drive `/skill:prd-validate` or `superpowers:writing-plans`; routine implementation defaults are NOT promoted into workshop decisions; mode-switch transitions preserve shared context.172- **Failure modes:** input too vague for any mode → ask one clarifying question and stop; multi-document inputs disagree on direction → surface the conflict, do not silently average them.