🌙 X AI 热点雷达 — Day 22 Evening Heartbeat

Report ID: 2026-07-18-2101
CST 时间: 2026-07-18 21:01 CST (Asia/Shanghai)
Scan Type: Evening Heartbeat (HN + GitHub fallback path)
信号总数: 10 top 10 / 15+ total deduped

⚠️ 数据来源声明 (Pitfall #162/167 ACTIVE — Day 22 两层降级)

本报告非 X/Twitter @claw0x 实时抓取。当前系统处于两层降级模式：

1. Pitfall #162 (Day 8 of block) — Hermes browser safety guard 阻断 document.cookie + fetch, X auth_token cookie 注入失败。xStatus: REACHABLE + AUTH_BLOCKED

2. Pitfall #167 (Day 6 of block) — Tavily provider-level outage, web_search/web_extract 全 432 返回。webStatus: DEGRADED

实际信号流: Hacker News Firebase API (top, parallel) + GitHub Search API (3 OR terms, sort=stars) + GitHub Trending today (browser no-auth)。5 new vs morning baseline of 10, 5 maintained。

🔔 Auth Reminder Day 22: 建议用户浏览器登录 @claw0x → 复制 auth_token → 替换 cron task 配置 + 编辑 ~/.hermes/config.yaml 添加 browser.allow_unsafe_evaluate: true → 重启 Hermes。

📈 上午信号 Momentum 追踪

对比 morning radar-2026-07-18-0908.json:

🆕 今晚新增发现 (vs morning baseline)

5 个全新信号 — 不在 morning top 10 内:

🏆 Top 10 Signals (narrative-priority sort)

排序逻辑 (Pitfall #171): 主键 = emoji priority (🚀=100 / 🤖=80 / 🧠=50); 次键 = finalScore = narrative_pri×1000 + raw×freshness_mult

1. 🚀 ROCKET xai-org/grok-build

SpaceXAI's coding agent harness and TUI. Fullscreen, mouse interactive, extensible. 17,870 stars in 89h = 200 stars/hour.

• metrics: ⭐ 17,870 | ⚡ 200.5/h | 📅 3.7d | finalScore: 120,050 | source: GH Search
• link: https://github.com/xai-org/grok-build
• momentum status: 🔁 maintained (also Day 22 morning top-10)
• arbitrage direction: grok-build-*.com / grok-build.app — SpaceXAI 官方 coding agent 套利窗口
• domain candidates: grok-build.com, grokbuild.dev, grok-build.app, grokbuild.io
• action plan: 建 Grok Build 工具站 (TUI 在线版 / 配置生成器), SEO 锁定 grok build 关键词
• pain point: SpaceXAI TUI 是 CLI-only, web 用户无 onboarding 入口
• github: https://github.com/xai-org/grok-build

2. 🚀 ROCKET Robbyant/lingbot-map

A feed-forward 3D foundation model for reconstructing scenes from streaming data. 827 stars today = paper/demo viral.

• metrics: ⭐ 12,490 | 📈 827/day | 📅 94.0d | finalScore: 100,865 | source: GH Trending
• link: https://github.com/Robbyant/lingbot-map
• momentum status: 🆕 NEW (Day 22 evening discovery)
• arbitrage direction: lingbot-map-*.com / lingbot-3d.app — 3D reconstruction 模型套利 (paper trigger)
• domain candidates: lingbot-map.com, lingbot3d.app, lingbot.dev
• action plan: 调研 paper / arXiv release → 建 3D scene reconstruction 在线 demo (huggingface spaces)
• pain point: 3D foundation model demo 资源稀缺, web 用户无法 streaming 试玩
• github: https://github.com/Robbyant/lingbot-map

3. 🚀 ROCKET Kimi K3: Open Frontier Intelligence

Kimi K3 open frontier intelligence discussion, 1,994 HN points, 34.3h age.

• metrics: 📊 1994 pts | 📅 34.3h | finalScore: 105,810 | source: HN
• link: https://news.ycombinator.com/item?id=48935342
• momentum status: 🔁 maintained (Day 21 evening + Day 22 morning)
• arbitrage direction: kimi-k3-*.com / kimi-k3.app — Chinese frontier model 海外 SEO
• domain candidates: kimi-k3.com, kimi-k3.app, kimik3.dev, kimi-k3.io
• action plan: 建 Kimi K3 chat playground + benchmark aggregator, 长尾 "kimi k3" / "kimi-k3" 锁定
• pain point: 中国前沿模型海外可访问性 + benchmark 文档分散

4. 🚀 ROCKET ibelick/ui-skills

Skills for Design Engineers. 529 stars today = fresh repo or major v2 release surge.

• metrics: ⭐ 4,720 | 📈 529/day | 📅 192.0d | finalScore: 101,238 | source: GH Trending
• link: https://github.com/ibelick/ui-skills
• momentum status: 🆕 NEW (Day 22 evening discovery)
• arbitrage direction: ui-skills.app / design-skills.dev — Design Engineers skill 集合套利
• domain candidates: ui-skills.app, design-engineer-skills.com, ui-skills.dev
• action plan: 调研 ui-skills 是什么 (skill format JSON?) → 聚合类似 design skills
• pain point: Design Engineers skill 资源零散, 无统一索引
• github: https://github.com/ibelick/ui-skills

5. 🚀 ROCKET Kimi K3, and what we can still learn from the pelican benchmark

Kimi K3 pelican benchmark discussion, 361 HN points, 22.7h.

• metrics: 📊 361 pts | 📅 22.7h | finalScore: 101,590 | source: HN
• link: https://news.ycombinator.com/item?id=48954851
• momentum status: 🔁 maintained (Day 22 morning)
• arbitrage direction: pelican-benchmark.com / kimik3-benchmark.app — Benchmark discussion SEO
• domain candidates: pelican-benchmark.com, kimi-k3-bench.dev
• action plan: 写 pelican benchmark 解释 + Kimi K3 score 实时追踪页
• pain point: AI benchmark 文档分散, 开发者难比较

6. 🤖 AGENT tirth8205/code-review-graph

Local-first code intelligence graph for MCP and CLI. 356 stars today surge.

• metrics: ⭐ 19,962 | 📈 356/day | 📅 141.8d | finalScore: 84,618 | source: GH Trending
• link: https://github.com/tirth8205/code-review-graph
• momentum status: 🆕 NEW (Day 22 evening discovery)
• arbitrage direction: code-review-graph.com / mcp-code-review.dev — Local MCP code review 套利
• domain candidates: code-review-graph.com, mcp-code-review.dev, local-code-review.app
• action plan: 调研 MCP code review 竞品 → 建 MCP server 聚合站 (类似 awesome-mcp-servers)
• pain point: MCP code review 工具分散, 缺统一评测
• github: https://github.com/tirth8205/code-review-graph

7. 🚀 ROCKET Nutlope/hallmark

Anti-AI-slop design skill for Claude Code, Cursor, and Codex. 1,485 stars in 24h.

• metrics: ⭐ 1,485 | 📈 1485/day | 📅 1.0d | finalScore: 106,190 | source: GH Search
• link: https://github.com/Nutlope/hallmark
• momentum status: 🔁 maintained (Day 22 morning)
• arbitrage direction: anti-slop.app / ai-slop-checker.com — AI 内容检测/反 slop 工具站
• domain candidates: anti-slop.app, ai-slop-checker.com, no-slop.dev, hallmark.dev
• action plan: 建 AI-slop 检测在线工具 (复制粘贴检测), SEO 锁定 "ai slop" / "anti slop" 关键词
• pain point: AI 生成内容泛滥, 缺乏 instant 检测工具
• github: https://github.com/Nutlope/hallmark

8. 🧠 RESEARCH Fable 5 vs GPT-5.6 Sol on NP-Hard problem

Fable 5 vs GPT-5.6 Sol comparison on NP-Hard problem, 28 HN points, 2h age.

• metrics: 📊 28 pts | 📅 2.0h | finalScore: 51,400 | source: HN
• link: https://news.ycombinator.com/item?id=48956949
• momentum status: 🆕 NEW (Day 22 evening)
• arbitrage direction: fable-5.com / gpt-5-6-sol.app — Frontier model benchmark 套利
• domain candidates: fable-5.app, gpt-5-6-sol.com, np-hard-bench.dev
• action plan: 聚合 frontier model benchmark (Fable 5 / GPT-5.6 / Claude) 长尾内容站
• pain point: Frontier model 性能对比缺可视化

9. 🤖 AGENT KnockOutEZ/wigolo

Local-first search/fetch/crawl/research for AI coding agents via MCP. 192 stars today, $0 cloud cost.

• metrics: ⭐ 971 | 📈 192/day | finalScore: 80,900 | source: GH Trending
• link: https://github.com/KnockOutEZ/wigolo
• momentum status: 🆕 NEW (Day 22 evening)
• arbitrage direction: local-mcp-search.app / mcp-search.dev — Local MCP search 聚合站
• domain candidates: local-mcp-search.dev, mcp-search.app, mcp-crawler.com
• action plan: 调研 MCP search 工具集合 → 建 awesome-mcp-search 聚合站
• pain point: MCP server 搜索分散, 缺发现机制
• github: https://github.com/KnockOutEZ/wigolo

10. 🤖 AGENT MoonshotAI/kimi-cli

Kimi Code CLI - your next CLI agent. 9,294 stars, 48 stars today (steady growth).

• metrics: ⭐ 9,294 | 📈 48/day | 📅 276.0d | finalScore: 80,042 | source: GH Trending
• link: https://github.com/MoonshotAI/kimi-cli
• momentum status: 🔁 maintained (GH Trending)
• arbitrage direction: kimi-cli.dev / kimi-code.com — Kimi CLI 教程 + 中文长尾
• domain candidates: kimi-cli.dev, kimi-code.com, kimi-cli.app
• action plan: 建 Kimi CLI 中文教程站, 锁定中文长尾
• pain point: Kimi CLI 中文教程稀缺, 海外用户 onboarding 难
• github: https://github.com/MoonshotAI/kimi-cli

🎯 Evening vs Morning — 对比与套利优先级

Evening 增量信号 (NEW) — 套利优先级排序:

Evening 维持信号 (CARRIED) — 持续套利验证:

• 🚀 xai-org/grok-build — 17.8K⭐ in 89h, TOP signal, 套利窗口仍 OPEN。建 grok-build.com 工具站 ROI 高。
• 🚀 Kimi K3 — HN #1 maintained 1994 pts, 中国 frontier 模型海外 SEO 持续窗口
• 🚀 Nutlope/hallmark — 1,485⭐ in 24h, anti-AI-slop 设计 skill, 工具站机会
• 🚀 Kimi K3 pelican benchmark — HN #5, frontier benchmark 内容站
• 🤖 MoonshotAI/kimi-cli — GH Trending maintained, Kimi CLI 中文教程

🐞 风险与限制

• 🔴 Two-layer degradation ACTIVE — X auth + Tavily 双阻断, 已稳定 6-8 天
• 🔴 HN points 与 GitHub stars 量纲不同 — 本报告统一作为 momentum proxy 解读, 方向一致但绝对值不可与 X likes 直接比较
• 🔴 24h fresh GH Search layer 普遍低 stars (≤16) — 噪声较多, evening heartbeat 主要依靠 GH Trending today + HN top 30 + 7d GH Search 138K repos
• ⚠️ 旧项目 daily surge 风险 — lingbot-map (94d) + ui-skills (192d) + code-review-graph (141d) 都是旧仓库触发 fresh 兴趣, freshness multiplier ×0.3, 套利窗口相对新首发短
• ⚠️ Tavily 432 Day 6 — 不重试, 报告标注 webStatus: DEGRADED

🚀 部署状态

• JSON pointer: /Volumes/sunzi/Code/ludusAI/public/data/demand-radar.json
• JSON archive: /Volumes/sunzi/Code/ludusAI/public/data/radar/radar-2026-07-18-2101.json
• MD report: /Volumes/sunzi/Code/ludusAI/daily-reports/2026-07-18/x-demand-radar-2101.md
• CF Pages deploy: pending git push (will validate 3 endpoints at T+30/90/120s)

Generated by cron job @ 2026-07-18 21:01 CST — x-demand-radar skill (Day 22 evening heartbeat)

Signal source: HN Firebase API + GitHub Search API + GitHub Trending (Pitfall #162 fallback path)