v1.8.0

Multi-sided awareness + cold-start brainstorming

Two new conditional capabilities — single-sided, non-network apps skip them entirely and stay just as lean.

Added

• Multi-sided / marketplace discovery awareness. A new Phase 0 "how many sides?" gate detects two-sided products, and the skill then discovers each side, not just the one the founder happens to be: per-side ICP scoring (the second side's basics are the first side's table stakes), the other side's struggling moment, both-sides flows, and a compound riskiest-assumption ("both sides actually show up," and which side is harder to get).
• Cold-start brainstorming (Phase 6.6). After finding the growth loop, the skill asks whether the loop can even start. For marketplaces / networks / social apps that need critical mass, it brainstorms a bootstrap strategy with the user — single-player mode first, start narrow, hold the network behind a liquidity threshold, seed the hard side by hand, seed supply not demand — and names a minimum-liquidity threshold to cross before opening the doors.
• references/MULTI-SIDED.md and references/COLD-START.md — the fuller playbooks, loaded only when relevant.

Prompted by a Mag7 PM's feedback that the ClearList example grilled the seller but never the buyer.

v1.7.2

Fix frontmatter YAML (block-scalar description) so the skill is installable via the skills CLI and indexable on skills.sh

v1.7.1

Add an ethical lens and a craft second look to Phase 2, and deepen Crazy 3s into comparable sketches

v1.7.0

Phase 6.6: Growth Loops (the engine that compounds)

vibe-check now helps a beginner find the one way their app recruits its next user on its own, preferably viral and organic, right after the Distribution phase.

Added

• Phase 6.6: Growth Loops. Reframes growth from a one-way funnel ("pour effort in the top forever") to a loop ("using the app creates the next user"), with three buildable shapes in plain language:
  • Content loop — public output found on Google (Reddit, recipe blogs, Substack)
  • Invite loop — using it pulls in someone new (Box, Figma, Google Docs)
  • Signal loop — others see it and copy (Livestrong bands, "Sent from my iPhone", Calendly links)
• Pushes building the loop into the core flow tied to the Phase 2 aha moment, names the one metric that proves it's spinning, and is honest that not every app has a loop — a faked one is worse than none.
• references/GROWTH-LOOPS.md — the fuller playbook: why a loop beats a funnel, the canonical examples (Netflix, LinkedIn, Uber, Substack, Airbnb), the full loop taxonomy, how to find and sketch your loop, and the four accelerators.
• A matching Growth Loop section in the plan document.

Credits

Brian Balfour / Reforge (growth loops), with Casey Winters and Kevin Kwok (loop taxonomy).

v1.6.0

De-bloat: dedupe SKILL.md under 500 lines, merge the three build references into MANAGING-YOUR-AI.md, trim the description under the 1024-char limit, fix portability

vibe-check v1.5.0

Opportunity-solution-tree borrows (Teresa Torres): evidence tags on the opportunity table, needs-in-user-language, a framing check and a riskiest-assumption test before building, and a sharper desired-outcome. See CHANGELOG.md.

vibe-check v1.4.0

Deeper, more faithful ODI: a competitor gap matrix (the stand-in for the satisfaction survey), ICP segmentation, the significantly-better rule. Plus a FrontierCode-inspired code-quality Definition of Done (fail-first tests, build/lint green, 'working is the floor'), and an optional side-by-side Crazy 3s comparison board. See CHANGELOG.md.

vibe-check v1.3.0

Adds Phase 6.5: Distribution, the final boss. The skill now forces a specific answer to who will reach the first users and how, reusing the discovery communities as the launch channel, instead of assuming build-it-and-they-will-come. See CHANGELOG.md.

vibe-check v1.2.0

Discovery now scores opportunity from two sources: Reddit for pain, and G2/Capterra/app-store reviews for how well current tools serve the need (feeding table stakes vs differentiator). Plus a willingness-to-pay check. Review sites use the same search-first fetch ladder as Reddit. See CHANGELOG.md.

vibe-check v1.1.0

Fixes discovery for tools that can't fetch Reddit directly (Claude Code included). Beat 2 now reaches Reddit via site:reddit.com web search, Reddit's read endpoints, or a manual hand-off, ranked by signal. See CHANGELOG.md.