Gap Finder

Finds viable ideas through adversarial generation-and-kill cycles. Generates batches of 5, validates every idea against real market data, kills failures with documented reasons, and loops until survivors emerge.

Modes

• Batch (default) — broad generate-and-kill across a domain. Generates batches of 5, runs the 6-check kill chain on each, loops until 3 survivors emerge. Best for "give me side-project ideas", "what should I build", scanning many adjacent niches. This is the workflow below.
• Deep (--deep) — exhaustive depth-first exploration of ONE space. Branches the market into opportunity tuples (market × wedge × unfair-advantage × why-now × who-pays), reality-calibrates each against real companies, maps named incumbents + GTM, runs an independent codex cross-attack and a mandatory steelman per node, then puts the top survivors through a VC-gauntlet fundability gate. Resumable via a JSON state file. Best for "find the real opportunities in X", "keep digging until you exhaust X". The full method lives in DEEP-DFS.md — read and follow it when --deep is passed, instead of the batch workflow below.

Workflow

1. Establish the domain — identify the space the user wants ideas in (e.g., "MCP servers", "developer tools", "mobile apps for nurses"). Extract any constraints (tech stack, budget, solo founder, etc.) and previously killed ideas. Write these to a scratchpad file in the current working directory (gap-finder-state.md).
2. Generate a batch of 5 ideas — each must be concrete and specific to the domain. Follow the idea template in GENERATION.md. Draw from at least 3 different angles per batch.
3. Run the kill chain on every idea — execute all 6 validation checks from VALIDATION.md. Search the web for each check. Kill on first fatal failure.
4. Record killed ideas — append every killed idea + specific kill reason to gap-finder-state.md. This prevents re-proposing.
5. Present survivors — if any ideas survive all 6 checks, present them with full evidence. See SURVIVOR-FORMAT.md.
6. Widen on saturation — if 2 consecutive batches have 0 survivors AND >80% died at Check 1 (competitors), the current sub-space is saturated. Do NOT stop. Instead: (a) note the saturated sub-space in state file, (b) expand the domain — go more niche, go adjacent, change the target user, change the business model, or combine two domains. Keep generating. The goal is survivors, not a saturation report.
7. Loop until 3 survivors — go to step 2. There is no batch limit. Keep generating until 3 survivors are found. After every 5 dead batches, radically pivot: change the domain, the target user, the business model, or the geography. The answer exists — you haven't looked hard enough yet.
8. Report survivors as they emerge — present each survivor immediately when it passes all 6 checks (don't wait for 3). After 3 survivors, write the final ranked summary.