with one click
idea-buildability
判断一个产品 idea 能否被 2-5 人小团队在 3 个月内做出 MVP。
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Menu
判断一个产品 idea 能否被 2-5 人小团队在 3 个月内做出 MVP。
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
根据本地外链 JSON 队列或批次队列,把当前项目发布到免费外链、目录、startup listing、AI tools 和 gallery 平台。用于需要 Codex 使用 @chrome 操作用户已登录的 Chrome、读取项目资料夹、按单个 JSON 或 30 条左右的 batch JSON 逐组处理、通过 subagent 调用 $humanizer 生成自然文案、填写平台提交表单、直接提交免费 listing 并更新队列状态的任务。
使用 Firecrawl 根据用户提供的项目落地页或产品 URL 抓取并保存本地项目资料文件夹。用于需要 Codex 获取网站摘要、页面 metadata、品牌 Logo、首屏截图和结构化 product 字段,并在当前项目下保存 profile.md、metadata.json、raw-firecrawl.json 和下载后的图片资产。
按 GoBackendFram 框架约定开发、配置、集成或审查 Go 后端功能.
Use for practical and strategic Google Ads work for any product: workflow diagnosis, first-launch readiness, landing page/material/copy prep, Keyword Planner usage, competitor/category research, Search campaign setup, ad groups, keywords, match types, responsive search ads, URL options/UTM, conversions, bidding, regions/languages, negatives, search-term analysis, daily optimization SOP, product/landing-page/onboarding/operations feedback loops, market expansion, growth experiments, and reviews from Google Ads CSVs, analytics screenshots, database funnels, or conversion screenshots. Use stored real Google Ads UI knowledge first; ask for screenshots/data only when needed. Never invent recommendations without evidence; separate evidence, inference, and hypotheses.
Use when asked to analyze a live website, SaaS product, web app, landing page, competitor site, or AI tool, especially with logged-in UX, Google/OAuth login, pre-authenticated Chrome sessions, payment avoidance, SEO/keyword research, competitor comparison, feasibility scoring, or a Markdown report saved locally.
Use when collecting today's products from Product Hunt, BetaList, Uneed, DevHunt, Microlaunch, Peerlist Launchpad, Fazier, Startup Fame, There Is An AI For That, Futurepedia, Toolify, AIToolHunt, or TopAI.tools into per-site JSON files.
| name | idea-buildability |
| description | 判断一个产品 idea 能否被 2-5 人小团队在 3 个月内做出 MVP。 |
读 input.md(包含 idea + 上下文 + evidence),严格判断**"2-5 人 3 个月能不能上线 MVP"**,并把判定写到 verdict.json。
工作目录下有一个 input.md 文件,内容由调用方写入,包含:
核心问题:"我接到这个 idea,2-5 人 3 个月能交付 MVP 吗?关键技术或数据资源现成可拿吗?"
行动:
web_search、curl 抓官方文档 / 定价 / TOS)。pass — 技术栈成熟、数据现成、工时可控。concern — 有难点但可绕过。fatal — 触发任意一条:
fatal_reason。核心问题:"用户愿付钱吗?PMF 在哪?半年内能 1000 付费用户吗?"
行动:
pass — 付费意愿明确、单边可冷启动、PMF 路径清晰。concern — 有不确定但可试。fatal — 触发任意一条:
fatal_reason。核心问题:"这种 idea 历史上谁试过为什么挂?现在做窗口还在吗?"
行动:
pass — 历史窗口仍开、无明显死亡名单。concern — 有死亡先例但有差异化。fatal — 触发任意一条:
fatal_reason。overall=reject,fatal_role 填首个 fatal 角色名。overall=pass。写到工作目录 verdict.json,只此一份,UTF-8,纯 JSON。Schema:
{
"overall": "pass | reject",
"fatal_role": "CTO | PM | FailedFounder | null",
"reasons": ["短句 fatal_reason", "..."],
"role_detail": {
"CTO": {
"verdict": "pass | fatal | concern",
"reasoning": "150 字推理",
"evidence_urls": ["https://...", "..."],
"fatal_reason": "一句话,verdict=fatal 时必填,否则 null"
},
"PM": { "...同上..." },
"FailedFounder": { "...同上..." }
}
}
reasons 数组:每个 fatal_role 的 fatal_reason 都收集进来(可有 0-3 条)。
web_search(query) — Tavily / 内置(优先,信息最新)curl -fsSL <url> — 抓官方文档 / 定价页 / TOSinput.md,写 verdict.json写完 verdict.json 立即结束。不要在 stdout 打印 verdict,不要生成其他文件,不要继续讨论。