# Ideation Engine > Ideation Engine - Skill: `handsomeboy990/ideation-engine` (Agent Skill, multi-file: 4 files) - Install (CLI): `npx skillmds@latest add handsomeboy990/ideation-engine` - Raw SKILL.md: https://api.skillmd.com/api/skills/handsomeboy990/ideation-engine/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: Handsomeboy990 (https://skillmd.com/u/handsomeboy990) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/handsomeboy990/ideation-engine --- # Ideation Engine Generating a hundred ideas is easy and nearly worthless: most are variations of three, and the volume hides the lack of range. This skill generates a smaller, genuinely diverse set, each grounded in a real need, that is worth the evaluation that follows. It is the discovery half of the opportunity method, done so the evaluation half has real material. ## 1. Frame the problem and audience first ``` problem the real problem to solve, stated as a problem, not a solution in disguise; "people forget to drink water" is a problem, "a water app" is a solution that has skipped the framing audience who actually has this problem, specifically; a problem no named audience has is not a problem why now what makes this worth solving now: a change, a gap, an unmet need constraint the actor's constraints from opportunity-core, so generation stays within what could actually be pursued ``` Ideas generated before the problem is framed solve whatever the generator found interesting, not what the audience needs. ## 2. Generate along deliberate axes A flat brainstorm clusters around the obvious. Deliberate axes force range. ``` axes vary the approach deliberately: different mechanisms, different audiences within the problem, different business models, different levels of ambition, the opposite of the obvious approach kind the ideas should differ in kind, not in wording; ten names for the same app is one idea, not ten extremes include the deliberately minimal and the deliberately ambitious version; the useful answer is often between them, and neither is seen without generating both inversion generate the approach that does the opposite of the default; it surfaces assumptions the default hides ``` ## 3. Ground each idea in a real need ``` premise each idea rests on a specific, real need of the named audience, stated as a hypothesis to test, per opportunity-core, not as a certainty distinguish an idea grounded in a real observed need from one grounded in a plausible-sounding assumption; label which is which evidence where a need is asserted, note what would confirm it, so evaluation and idea-evaluation can test it rather than assume it ``` ## 4. Hand over a diverse, non-redundant set ``` diverse the set spans the axes, so evaluation chooses between real alternatives, not shades of one few a handful of genuinely different ideas beats a hundred variations; the count is not the metric, the range is non-redundant collapse the near-duplicates before handing over; ten framings of one idea waste the evaluation ready each idea carries its problem, audience, premise, and the axis it came from, so evaluation has what it needs ``` ## 5. Prohibitions - Never generate before framing the problem and naming the audience. - Never mistake a solution for a problem in the framing. - Never produce a long list of variations and call it a diverse set. - Never present a plausible-sounding assumption as an established need. - Never let volume substitute for range; ten near-duplicates are one idea. - Never generate ideas that ignore the actor's real constraints. - Never skip the extreme and inverted versions; the obvious middle is not the whole space. ## 6. Protocol 1. Frame the real problem, the specific audience, why now, and the constraints. 2. Choose deliberate generation axes to force range. 3. Generate along the axes, including the minimal, the ambitious, and the inverted. 4. Ground each idea in a specific real need, marked as observed or hypothesised. 5. Collapse near-duplicates; keep the genuinely different. 6. Hand over the diverse set, each with its problem, audience, premise, and axis. 7. Pass to `idea-evaluation` for the convergent half. ## 7. Auto-critique Score from 0 to 5: problem framed as a problem with a named audience before generating, axes deliberate and forcing range, extremes and inversion included, each idea grounded and marked observed or hypothesised, near-duplicates collapsed, the handed-over set genuinely diverse, constraints respected. Threshold: no axis below 3, average at least 4. A long list of variations presented as diversity, or an assumption presented as an established need, caps the score until fixed, per opportunity-core. ## 8. Interfaces - Upstream: `opportunity-core` for the method and grounding rule. - Downstream: `idea-evaluation` converges the set to a ranked few; `hackathon-strategy` uses it to generate project ideas for a challenge; `market-research` and `client-discovery` when the ideas are business directions. - Lateral: `project-brief` when a chosen idea becomes a project; `technology-selection` when it becomes a build. ## 9. Domain independence The engine works the same for a software feature, a startup, a research direction, a piece of content, a hackathon project, or a personal project. What changes is the axes: business models for a startup, mechanisms for a feature, angles for content, methods for research. The framing-first, range-over-volume, grounded- premise discipline is constant across all of them. ## 10. Note on the count The instinct to measure ideation by the number of ideas is the thing to resist. A set of six ideas that differ in kind, each grounded and ready to evaluate, is a better output than sixty that cluster around three premises. When asked for "lots of ideas", generate for range and hand over the diverse few, with a note that quantity was traded for the diversity that makes evaluation meaningful.