| name | hackson-ideate |
| description | Generate and filter hackathon ideas against judging criteria. Three rounds — diverge, converge, lock in. Use after selecting a competition. |
Hackson Ideate — Idea Generation & Selection
You help the user generate winning ideas for their selected hackathon. Three rounds: diverge (many ideas), converge (deep analysis of favorites), lock in (final spec).
Prerequisites
Read the current competition's hackson-state.md (under competitions/<name>/) to get:
- Competition name, theme, judging criteria, tech requirements, deadline
- If no active competition or phase isn't
ideating, tell user to run /hackson-scout first.
Round 1: Diverge (8-10 ideas)
Generate 8-10 ideas. For each one:
Idea N: [Name]
- One-liner: What it does in one sentence
- Differentiator: Why this stands out from obvious ideas
- Effort: X hours (realistic for solo dev with AI assistance)
- Stack: Key tech choices
- Judges would like it because: [map to specific judging criteria]
Generation strategy:
- 3-4 ideas that directly address the stated theme (safe bets)
- 2-3 ideas that creatively reinterpret the theme (differentiated)
- 2-3 ideas from adjacent domains that technically qualify (unexpected angle)
- Reference patterns from past hackathon winners: developer tools, visualization dashboards, novel integrations, social impact angles tend to score well
Present all ideas, then ask: "Which 2-3 interest you? Or want me to generate more in a specific direction?"
Round 2: Converge (Deep Analysis)
For each selected idea, provide:
[Idea Name] — Deep Dive
Judge Scorecard (predicted):
- Innovation: X/10 — [why]
- Completeness: X/10 — [what MVP achieves in available time]
- Utility: X/10 — [real-world value argument]
- Technical Depth: X/10 — [what's impressive under the hood]
48-hour MVP scope:
- Must-have: [2-3 features that make the demo work]
- Nice-to-have: [1-2 features if time allows]
- Cut: [things that sound cool but aren't worth the time]
Biggest risk:
- [Technical blocker, API dependency, scope creep vector, etc.]
- Mitigation: [how to handle if it happens]
vs. Competition:
- Others will likely build: [obvious approaches]
- This stands out because: [differentiation argument]
Then ask: "Which one do you want to go with? Or want me to dig deeper on any?"
Round 3: Lock In
User picks one idea. Generate the Idea Spec:
Idea Spec
Name: [Project name]
One-liner: [Elevator pitch]
Target User: [Who uses this]
Core Features:
- [Feature 1 — most important]
- [Feature 2]
- [Feature 3 — stretch]
Tech Stack:
- Frontend: [choice + why]
- Backend: [choice + why]
- Deploy: [choice + why]
- Key APIs/libraries: [list]
Demo Vision:
[What the demo video shows — the "wow moment" in 30 seconds]
Success Criteria:
[What "done" looks like for submission]
Write this to the ## Idea section of the competition's hackson-state.md. Set phase to building.
Rollback Entry Point
If called when phase is already building (user is coming back from a failed build):
- Acknowledge: "Build didn't work out. Let's find a better idea."
- Ask: pivot (similar direction, different approach) or restart (full diverge)?
- Preserve competition info, clear idea section.
