研究内容为英文. 本页的研究内容（标题、简报、来源）以英文撰写，未经机器翻译。只有界面已翻译。

新想法AI app builders

Craft Anything-powered "Recreate this" hosted…

1周前 · 评分于 2026年11月4日 · ID HBV-261104-01

暂定标题： Craft Anything-powered "Recreate this" hosted service (clean-room only) — working name "Remake Studio"

摘要

简报已验证假设估算

Benchmark: storytold/craft-anything (ArtCraft team): 10,364 stars in 4 days, +4,269 on 3 Nov (peak, last full day), MIT OR Apache-2.0, v0.3.0 released 2 Nov 2026 (Verified). Study → Build → Check → Refine loop with 300+ Rust inspection tools; clean-room mode reads only the visible surface; Deep mode reads code for private use.…

阅读其余部分

Existence check: No hosted Craft Anything service found (maintainer roadmap item R4.2 "Managed tier" is open, "not before M5"; Verified docs/ROADMAP.md). Generic hosted builders that accept "clone this URL" prompts exist (Lovable, Bolt, v0, Same) — Hypothesis on how close their output is; none runs a measured rebuild-and-compare loop that we found. MIT or Apache-2.0; no hosted version from the maintainer yet, a managed tier is only on its roadmap; X 1,301 posts/7d rising). The wrapper idea scores POS 57, not above 70: the cost of one Study→Build→Check run is unknown and could be several dollars at API prices (no verified figure exists); server-side building with an API key is not yet live in the engine; cloning third-party sites carries IP and terms-of-use risk; and general builders (Lovable, Bolt, v0, Same) already accept "copy this site" prompts. Audience (makers and developers) and desktop/local design are recorded as risks only, never as cuts (Cesar, 4 Nov 09:36 BRT). Stars: Total 10,364 (github.com HTML, 4 Nov ~12:00 BRT; API 10,362). Created 31 Oct 2026, MIT OR Apache-2.0. 24h (3 Nov) +4,269; 7d +8,305 ≈1,186/day; since creation ≈2,590/day; 7v7 and 3v3 not computable (repo 4 days old), daily series 990 → 3,046 → 4,269 accelerating; peak 3 Nov (last full day), not past peak; today partial +2,078 (OSS Insight). Nothing estimated. X: 1,301 posts/7d (Rising +2700% 3v3 (1–3 Nov 1,120 vs 29–31 Oct 40; daily full UTC days 18, 6, 16, 97, 536, 487); peak 2 Nov 536; 3 Nov is 9% below peak). Revenue hypothesis: ~$2K–$10K/month at month 6 (HYPOTHESIS; depends on the unmeasured cost per run).

描述假设模型

Paste the address of a website or web app and get your own clean-room rebuild (new code, your brand) as a running preview, using the open-source Craft Anything engine on our servers — no Rust build or Claude Code account needed. Who: Indie makers, founders, agencies and product teams who want a working starting point "like that site, but ours" (also usable inside HBV for copycat MVPs and as a feature for Startup Ideas HQ).…

阅读其余部分

Pain: Craft Anything needs a local install plus the user’s own signed-in coding agent and burns their plan quota; Monetization: Credits per rebuild (each credit covers the model cost of one Study→Build→Check run plus margin), subscription for agencies; free first page as viral hook (Hypothesis; no approved price). plus a file-writing loop if Claude Code headless with an API key is not used; stack: Craft Anything CLI/MCP server on Linux workers (Fly/Hetzner), Claude API under Anthropic Commercial Terms (our key), headless browser, Vercel previews, Stripe. GTM: "Recreate <site> as open source" SEO/GEO pages, share-your-rebuild gallery with likeness score (viral loop), maker communities; faceless demos.

目标客户

Indie makers, founders, agencies and product teams who want a working starting point "like that site, but ours" (also usable inside HBV for copycat MVPs and as a feature for Startup Ideas HQ).

痛点

Craft Anything needs a local install plus the user’s own signed-in coding agent and burns their plan quota

为何是这个分数

分数较低的原因： Unknown and possibly high model cost per run (no verified figure) combined with IP/terms-of-use risk of cloning other people’s sites; generic builders already do "copy this site".

护城河假设

Weak to moderate: Measured likeness score and a public gallery of rebuilds create SEO pages and a share loop. Operational know-how running a heavy Rust engine and coding agents cheaply at scale. Status: Hypothesis.

需求信号

X 上的帖子（7 天）

1301

信号趋势

Rising +2700% 3v3 (1–3 Nov 1,120 vs 29–31 Oct 40; daily full UTC days 18, 6, 16, 97, 536, 487); peak 2 Nov 536; 3 Nov is 9% below peak

网络需求

Press coverage 1–2 Nov 2026: madrobot.blog, fervorai.dev, doppingcastle.com (read via web search 4 Nov); GitHub v0.3.0 release 2 Nov; open issues show heavy use (e.g. #731 token over-consumption report, 3 Nov).

市场与竞争对手

TAM

Not sized this run — Scope: Global. Method: No third-party report fetched. (Basis: n/a)

SAM假设

Not sized — Scope: Target markets above. Method: HBV assumption. (Basis: Hypothesis)

SOM假设

~$24K–$120K ARR (Hypothesis) — Scope: Month-6 run-rate. Method: From revenue hypothesis ×12. (Basis: Hypothesis)

竞争对手

Lovable, Bolt.new, v0, Same (hosted "copy this site" builders); Craft Anything itself free on desktop; screenshot-to-code repos. HBV angle: measured likeness score and clean-room rebuild as a service.

商业模式

变现方式假设模型

Credits per rebuild (each credit covers the model cost of one Study→Build→Check run plus margin), subscription for agencies; free first page as viral hook (Hypothesis; no approved price).

收入假设（每月）假设

~$2K–$10K/month at month 6 (HYPOTHESIS; depends on the unmeasured cost per run).

待完成的任务

When I see a product page or small web app I admire, I want to get a working, rebranded rebuild in my own stack in minutes, so I can start from a proven layout instead of a blank page.

HXC

Diego, 31, solo founder in Mexico City (illustrative) · Ships landing pages and small SaaS with Cursor; studies competitors’ sites weekly. Buying signal: Already pays for Lovable or v0 and a Claude plan.

构建与 GTM

技术栈

Craft Anything CLI/MCP server on Linux workers (Fly/Hetzner), Claude API under Anthropic Commercial Terms (our key), headless browser, Vercel previews, Stripe

GTM 渠道

"Recreate <site> as open source" SEO/GEO pages, share-your-rebuild gallery with likeness score (viral loop), maker communities; faceless demos

风险

主要风险

Unknown and possibly high model cost per run (no verified figure) combined with IP/terms-of-use risk of cloning other people’s sites; generic builders already do "copy this site".

全部风险已验证

Value: Supported — 10,364 stars in 4 days; X rising (Verified) | Usability: Attention — Users may expect pixel-perfect clones | Feasibility: High — Server-side build with API key not live in the engine | Business Viability: High — Unknown cost per run; clone IP/terms risk

下一步验证

Run one clean-room build of a sample page with an API key and read the engine’s own usage.json (tokens, estimated $) to get a Verified cost per run; re-read stars and X on 6 Nov.

来源

newgithub-wrappergithubai-builderclean-roomdeveloper-toolsglobal

想在评分当天收到下一个想法吗？

分数、来源、风险和下一步，通过邮件发送。

ProductBots.ai

喜欢这个想法吗？用 ProductBots.ai 把它变成计划。

ProductBots.ai 是一个 Agentic Product OS：由智能体与你一起完成产品工作，从发现到计划。下载这个想法的简报，作为起点。

打开 ProductBots.ai

简报文件为英文。

预览简报

# Briefing: Craft Anything-powered "Recreate this" hosted service (clean-room only) — working name "Remake Studio"

Source: Startup Ideas HQ (HB Ventures), found 2026-11-04. ProductBots Opportunity Score: 57/100, evidence confidence 54%.
Evidence labels (Verified, Self-reported, Hypothesis) are kept as written by the researcher.

## Goal for ProductBots.ai
Turn this idea into a validated MVP plan and a first build. Start from the next validation step; do not assume unlabelled numbers are verified.

## What it is
Paste the address of a website or web app and get your own clean-room rebuild (new code, your brand) as a running preview, using the open-source Craft Anything engine on our servers — no Rust build or Claude Code account needed.

Benchmark: storytold/craft-anything (ArtCraft team): 10,364 stars in 4 days, +4,269 on 3 Nov (peak, last full day), MIT OR Apache-2.0, v0.3.0 released 2 Nov 2026 (Verified). Study → Build → Check → Refine loop with 300+ Rust inspection tools; clean-room mode reads only the visible surface; Deep mode reads code for private use. Existence check: No hosted Craft Anything service found (maintainer roadmap item R4.2 "Managed tier" is open, "not before M5"; Verified docs/ROADMAP.md). Generic hosted builders that accept "clone this URL" prompts exist (Lovable, Bolt, v0, Same) — Hypothesis on how close their output is; none runs a measured rebuild-and-compare loop that we found. MIT or Apache-2.0; no hosted version from the maintainer yet, a managed tier is only on its roadmap; X 1,301 posts/7d rising). The wrapper idea scores POS 57, not above 70: the cost of one Study→Build→Check run is unknown and could be several dollars at API prices (no verified figure exists); server-side building with an API key is not yet live in the engine; cloning third-party sites carries IP and terms-of-use risk; and general builders (Lovable, Bolt, v0, Same) already accept "copy this site" prompts. Audience (makers and developers) and desktop/local design are recorded as risks only, never as cuts (Cesar, 4 Nov 09:36 BRT). Stars: Total 10,364 (github.com HTML, 4 Nov ~12:00 BRT; API 10,362). Created 31 Oct 2026, MIT OR Apache-2.0. 24h (3 Nov) +4,269; 7d +8,305 ≈1,186/day; since creation ≈2,590/day; 7v7 and 3v3 not computable (repo 4 days old), daily series 990 → 3,046 → 4,269 accelerating; peak 3 Nov (last full day), not past peak; today partial +2,078 (OSS Insight). Nothing estimated. X: 1,301 posts/7d (Rising +2700% 3v3 (1–3 Nov 1,120 vs 29–31 Oct 40; daily full UTC days 18, 6, 16, 97, 536, 487); peak 2 Nov 536; 3 Nov is 9% below peak). Revenue hypothesis: ~$2K–$10K/month at month 6 (HYPOTHESIS; depends on the unmeasured cost per run).

## Customer and pain
Target customer: Indie makers, founders, agencies and product teams who want a working starting point "like that site, but ours" (also usable inside HBV for copycat MVPs and as a feature for Startup Ideas HQ).
Pain: Craft Anything needs a local install plus the user’s own signed-in coding agent and burns their plan quota

## Market and competitors
TAM: Not sized this run — Scope: Global. Method: No third-party report fetched. (Basis: n/a)
SAM: Not sized — Scope: Target markets above. Method: HBV assumption. (Basis: Hypothesis)
SOM: ~$24K–$120K ARR (Hypothesis) — Scope: Month-6 run-rate. Method: From revenue hypothesis ×12. (Basis: Hypothesis)
Competitors: Lovable, Bolt.new, v0, Same (hosted "copy this site" builders); Craft Anything itself free on desktop; screenshot-to-code repos. HBV angle: measured likeness score and clean-room rebuild as a service.
Geography: Global English first; Portuguese/Spanish UI for LatAm later

## Business model
Monetization: Credits per rebuild (each credit covers the model cost of one Study→Build→Check run plus margin), subscription for agencies; free first page as viral hook (Hypothesis; no approved price).
Revenue hypothesis (monthly): ~$2K–$10K/month at month 6 (HYPOTHESIS; depends on the unmeasured cost per run).

## Build and go-to-market
Build effort (agent-weeks): ~3–5 agent-weeks (headless CLI on a workersandbox, preview hosting, billing); plus a file-writing loop if Claude Code headless with an API key is not used
Time to first revenue: ~6–8 weeks
Tech stack: Craft Anything CLI/MCP server on Linux workers (Fly/Hetzner), Claude API under Anthropic Commercial Terms (our key), headless browser, Vercel previews, Stripe
GTM channels: "Recreate <site> as open source" SEO/GEO pages, share-your-rebuild gallery with likeness score (viral loop), maker communities; faceless demos

## Risks and next step
Main risk: Unknown and possibly high model cost per run (no verified figure) combined with IP/terms-of-use risk of cloning other people’s sites; generic builders already do "copy this site".
Next validation action: Run one clean-room build of a sample page with an API key and read the engine’s own usage.json (tokens, estimated $) to get a Verified cost per run; re-read stars and X on 6 Nov.

## Sources
https://github.com/storytold/craft-anything; https://github.com/storytold/craft-anything/blob/master/docs/claude-code.md; https://github.com/storytold/craft-anything/blob/master/NOTICE; https://github.com/storytold/craft-anything/blob/master/docs/brand/LICENSE-brand.txt; https://github.com/storytold/craft-anything/blob/master/docs/ROADMAP.md; https://github.com/storytold/craft-anything/pull/9; https://github.com/storytold/craft-anything/issues/731; https://github.com/storytold/craft-anything/issues/272; https://madrobot.blog/2026/11/02/craft-anything-clone-any-website-app-game/; https://fervorai.dev/articles/craft-anything-turns-your-coding-agent-into-a-rebuild-and-compare-loop/; OSS Insight; X API counts

这是研究，不是投资建议。数字保留其发布时的标签。

摘要

简报✓已验证○假设≈估算

描述○假设≈模型