| name | idea |
| description | Generate and score 3 innovative hackathon project ideas aligned with the hackathon context |
| metadata | {"openclaw":{"always":false}} |
Idea Generation Skill
Purpose
Generate 3 creative, feasible project ideas for a hackathon.
Score each idea on multiple dimensions and recommend the best one.
When to Use
After hackathon research is complete (hackathon/context.json exists).
CRITICAL: Execute Directly — DO NOT use spawn()
Do NOT call spawn() for this task.
Execute every step yourself in this conversation, then write the output file.
Spawning a sub-agent will exhaust the tool budget on unnecessary searches and the file will never be written.
Also: DO NOT call web_search — it is not configured. All hackathon context you need is in context.json.
Direct Execution Steps
Step 1 — Read context:
read_file("hackathon/context.json")
Step 2 — Analyse the technical landscape from context.json:
- Identify the strongest APIs/platforms available
- List the most useful capabilities for a strong demo
- Prefer ideas with clear implementation paths and visible value
Step 3 — Generate 3 distinct ideas across these archetypes:
- Idea A — "AI Infrastructure": A platform/protocol other agents can use
- Idea B — "AI Application": A user-facing tool that solves a real problem autonomously
- Idea C — "Web3 x AI Hybrid": Combines on-chain mechanics with AI intelligence
For each idea:
- Core technologies are integral to the mechanism (not superficial add-ons)
- Problem is real and well-defined
- MVP is achievable in 5 days of coding
- Demo moment is clear and visual
Step 4 — Score each idea on these dimensions (1–5 each):
- innovation: how novel is the concept? (5 = never seen before)
- feasibility: can it be built in 7 days solo? (5 = straightforward)
- sponsor_depth: how integral are sponsor APIs? (5 = can't work without them)
- demo_impact: how impressive is the live demo? (5 = judges will remember it)
- market_fit: does it solve a real problem people care about? (5 = obvious pain point)
composite = (innovation×1.5 + feasibility×2.0 + sponsor_depth×2.0 + demo_impact×2.0 + market_fit×1.5) / 9.0
Step 5 — Write output with write_file("hackathon/ideas.json", ...):
{
"ideas": [
{
"id": "idea_a",
"archetype": "infrastructure|application|web3_ai",
"name": "string",
"tagline": "one compelling sentence",
"problem": "what pain point this solves (2-3 sentences)",
"solution": "how it works at a high level (3-4 sentences)",
"tech_stack": {
"backend": "string",
"frontend": "string or null",
"ai_models": ["string"],
"blockchain": "string or null",
"storage": "string"
},
"integrations": ["string"],
"architecture_sketch": "ASCII text diagram of key components",
"mvp_scope": "exactly what can be demoed in 7 days",
"wow_factor": "the one thing that makes judges say 'I've never seen this'",
"risks": ["top 3 risks"],
"scores": {
"innovation": 4,
"feasibility": 3,
"sponsor_depth": 5,
"demo_impact": 4,
"market_fit": 3,
"composite": 3.89
}
}
],
"recommendation": "idea_X",
"recommendation_rationale": "2-3 sentences explaining the choice",
"generated_at": "ISO timestamp"
}
Your task is complete once hackathon/ideas.json is written. Do not proceed to any other phase.
Output File