[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-93442":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":10,"languages":10,"totalLinesOfCode":10,"stars":11,"forks":12,"watchers":13,"openIssues":14,"contributorsCount":14,"subscribersCount":14,"size":14,"stars1d":15,"stars7d":16,"stars30d":16,"stars90d":14,"forks30d":14,"starsTrendScore":17,"compositeScore":18,"rankGlobal":10,"rankLanguage":10,"license":19,"archived":20,"fork":20,"defaultBranch":21,"hasWiki":20,"hasPages":20,"topics":22,"createdAt":10,"pushedAt":10,"updatedAt":23,"readmeContent":24,"aiSummary":25,"trendingCount":14,"starSnapshotCount":14,"syncStatus":26,"lastSyncTime":27,"discoverSource":28},93442,"whathappened","kunchenguid\u002Fwhathappened","kunchenguid","Agent skill: adaptive X-only briefing of what happened + public opinion + debates","https:\u002F\u002Fgithub.com\u002Fkunchenguid\u002Fwhathappened",null,197,13,1,0,4,69,31,78.34,"MIT License",false,"main",[],"2026-07-22 04:02:09","\u003Ch1 align=\"center\">\u002Fwhathappened\u003C\u002Fh1>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fdocs.x.ai\u002Fbuild\u002Foverview\"\n    >\u003Cimg\n      alt=\"Grok Build only\"\n      src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FGrok%20Build-only-black?style=flat-square\"\n  \u002F>\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fagentskills.io\"\n    >\u003Cimg\n      alt=\"Agent Skills\"\n      src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FAgent%20Skills-package-blue?style=flat-square\"\n  \u002F>\u003C\u002Fa>\n  \u003Ca href=\"LICENSE\"\n    >\u003Cimg\n      alt=\"License\"\n      src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Flicense-MIT-green?style=flat-square\"\n  \u002F>\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fx.com\u002Fkunchenguid\"\n    >\u003Cimg\n      alt=\"X\"\n      src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FX-@kunchenguid-black?style=flat-square\"\n  \u002F>\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fdiscord.gg\u002FWsy2NpnZDu\"\n    >\u003Cimg\n      alt=\"Discord\"\n      src=\"https:\u002F\u002Fimg.shields.io\u002Fdiscord\u002F1439901831038763092?style=flat-square&label=discord\"\n  \u002F>\u003C\u002Fa>\n\u003C\u002Fp>\n\n\u003Ch3 align=\"center\">What just happened on X - and what is the room actually saying?\u003C\u002Fh3>\n\n\u003Cp align=\"center\">\u003Cstrong>Grok Build only.\u003C\u002Fstrong> This skill needs Grok’s native X tools. Other agents can install the package; they cannot run it for real.\u003C\u002Fp>\n\nSomething drops. A model ships. A founder posts. A product melts down.\n\nYou open X and get dunks, screenshots, quote-tweet chains, and three conflicting \"official\" takes - none of them in one place.\n\n**whathappened** is a [Grok Build](https:\u002F\u002Fdocs.x.ai\u002Fbuild\u002Foverview) [Agent Skill](https:\u002F\u002Fagentskills.io) that turns that firehose into a short neutral briefing: what happened, where the conversation is, the opinion map, the live debates, and the receipts.\n\n- **X-first** - public conversation from X, not a blog roundup. One optional web lookup only to resolve *who\u002Fwhat* the topic is.\n- **Adaptive window** - breaking stories use minutes or hours; quieter ones widen only when the story is incomplete. Freshness wins by default.\n- **Opinion map + debates** - camps, rough sample shares, steelman vs critique, with real post links - not a vibes paragraph.\n\n## Quick Start\n\n```sh\n# install into Grok Build (global recommended)\n$ npx skills add kunchenguid\u002Fwhathappened -g\n# or: copy skills\u002Fwhathappened → ~\u002F.grok\u002Fskills\u002Fwhathappened\n\n# in Grok Build\n\u002Fwhathappened Kimi K3\n```\n\nExample output from a real run on launch day (edited only for README length; structure is what the skill produces):\n\n```markdown\n# \u002Fwhathappened: Kimi K3\n\n**Window:** last ~24h · mode Same-day · as of 2026-07-16 ~22:30 UTC\n**X sample:** ~60+ posts · Top + Latest + Semantic + first-party · 4 threads · confidence high (event) \u002F medium-high (opinion; day-0)\n**Entity resolve:** none (X user search → official @Kimi_Moonshot)\n\n## What happened\n\nMoonshot’s Kimi account launched Kimi K3 today as “Open Frontier Intelligence”: a 2.8T-parameter, 1M-context, native multimodal model for long-horizon agentic coding, live on Kimi product surfaces and API, with full weights promised by July 27, 2026.\n\nThe official thread also stresses architecture (Kimi Delta Attention, Attention Residuals, sparse MoE) and demos (kernel optimization, vision-in-the-loop game\u002FUI work).\n\nIndependent and community scoreboards moved quickly: Arena Frontend Code Arena #1 (over Claude Fable 5), Text Arena roughly top-10, and Artificial Analysis Intelligence Index ~57 (near Opus 4.8 \u002F GPT-5.5, behind Fable 5 and GPT-5.6 Sol).\n\n## Where the conversation is\n\n- Dominant frame: open weights (or soon-to-be) at near-frontier quality, especially coding\u002Ffrontend - often cast as a milestone for open models.\n- Secondary frames: API price\u002Flatency vs closed flagships; China vs US competitive pressure; policy\u002Fsafety backlash expected next.\n- Who is loud: Moonshot first-party; eval accounts (Arena, Artificial Analysis); coding-agent vendors (e.g. Cline); AI Twitter builders; a thinner but clear skeptic lane on cost\u002Fspeed\u002Freal-work gaps.\n\n## Public opinion map\n\nShares are rough qualitative judgment from this X sample, not polling.\n\n| Camp | Share (rough) | Core claim | Representative voices |\n|------|---------------|------------|------------------------|\n| Open-frontier bulls | ~50% | Near Fable\u002FSol on many axes; #1 frontend; open weights change the game | @cline, @arena, @TheAhmadOsman |\n| Nuanced practitioners | ~25% | Very strong (esp. frontend), not undefeated overall; slow \u002F thinks hard \u002F API not “cheap open” | @goodworse, @DimitriGilbert, @runzhuotao, @jaminball |\n| Geopolitics \u002F policy | ~15% | “Sputnik” for the US; fear of open-model bans; cycle of China\u002Fsafety rhetoric | @MTSlive (quoting Amp CEO), @Dan_Jeffries1, @haider1 |\n| Maximal hype | ~10% | AGI-adjacent language, “mogs everything” | e.g. @dedene |\n\n## The live debates\n\n1. **Is it frontier, or “almost frontier”?**\n   Side A: same ballpark as GPT-5.6 \u002F Fable 5, #1 frontend. Side B: AA puts it behind Fable\u002FSol overall; some real-workload tests put it nearer Opus 4.8 \u002F GPT-5.5.\n\n2. **Cheap open win vs expensive reasoning tax**\n   Side A: $3\u002F$15 per MTok framed as Sonnet-like value. Side B: “thinks a lot → costs a lot,” slower than hoped, pricier than prior Kimi models.\n\n3. **Competition catalyst vs regulatory panic**\n   Side A: healthy race, progress accelerates. Side B: expect essays, China-hawk policy, open-weight restrictions.\n\n4. **“Open” today vs weights on July 27**\n   API\u002Fproduct is live; weights are scheduled, not yet downloadable.\n\n## Notable posts (receipts)\n\n- @Kimi_Moonshot - origin: official launch specs + July 27 weights\n- @arena - scoreboard: Frontend Code Arena #1 over Fable 5\n- @ArtificialAnlys - independent eval: AA ~57; agentic gains; cost\u002Ftask framing\n- @cline - builder frame: largest open-weight class model, near Fable\u002FGPT-5.6, Sonnet-like API price\n- @synthwavedd - receipt: “drops at 57 on Artificial Analysis, just behind Fable and Sol”\n- @goodworse - steelman critique: frontend strength vs real-world closer to Opus 4.8\u002FGPT-5.5; slow; pricier than K2.7\n- @DimitriGilbert - hands-on: “really good… only downside, it thinks a LOT… and therefore costs a LOT”\n- @theo - ecosystem: practical “which harness?” question, not pure scoreboard talk\n\n## Gaps \u002F caveats\n\n- Day-0 recency bias: likes and quotes still forming; Top lane is announcement-heavy.\n- Weights not out yet: open-weight celebration is partly forward-looking (July 27).\n- English\u002FAI-Twitter skew; capacity\u002Fnoise on Latest.\n- What would change the read: multi-day coding failures, weights delayed, or closed labs shipping clear counters.\n```\n\n## Grok Build only\n\nThis skill is **not** a multi-harness research product. It depends on Grok Build’s native X stack:\n\n- `x_keyword_search`\n- `x_semantic_search`\n- `x_thread_fetch`\n- `x_user_search`\n\nIf those tools are missing, the skill should **refuse** rather than fake an X briefing from web search. Installing into Claude Code, Cursor, Codex, etc. will not give you a working run.\n\n## Install\n\n**Global (recommended)** for Grok:\n\n```sh\nnpx skills add kunchenguid\u002Fwhathappened -g\n```\n\n**Project-local:**\n\n```sh\nnpx skills add kunchenguid\u002Fwhathappened\n```\n\n**Direct into Grok’s user skills dir:**\n\n```sh\ngit clone https:\u002F\u002Fgithub.com\u002Fkunchenguid\u002Fwhathappened.git\ncp -R whathappened\u002Fskills\u002Fwhathappened ~\u002F.grok\u002Fskills\u002Fwhathappened\n# or: ln -s \"$(pwd)\u002Fwhathappened\u002Fskills\u002Fwhathappened\" ~\u002F.grok\u002Fskills\u002Fwhathappened\n```\n\nGrok also discovers skills under `.agents\u002Fskills\u002F`, so a normal `npx skills add -g` install often lands where Grok can see it. Prefer `~\u002F.grok\u002Fskills\u002F` when you want the Grok-native path explicitly.\n\n**List skills in this package without installing:**\n\n```sh\nnpx skills add kunchenguid\u002Fwhathappened --list\n```\n\n## How It Works\n\n```\ntopic\n  │\n  ▼\n┌─────────────────┐\n│ optional web x1 │  resolve entity only (name \u002F handle)\n└────────┬────────┘\n         ▼\n┌─────────────────┐\n│ pulse (Latest)  │  measure velocity → commit window mode\n└────────┬────────┘\n         ▼\n┌─────────────────┐\n│ search lattice  │  Top · Latest · Semantic · from: · debate\n└────────┬────────┘\n         ▼\n┌─────────────────┐\n│ thread fetch    │  origin · official · camps · steelman\n└────────┬────────┘\n         ▼\n┌─────────────────┐\n│ rank + cluster  │  engagement × freshness × on-entity\n└────────┬────────┘\n         ▼\n      briefing\n```\n\n- **Modes:** Breaking (hours) → Same-day → Story (days) → Background (longer only if needed) - shortest window that still explains the story.\n- **Soft web rule:** at most one lookup for identity. Sentiment stays X-only.\n- **No discovery mode in v1:** you must name a topic (`\u002Fwhathappened` alone asks for one).\n\n## Usage\n\n| Invoke | Example |\n| ------ | ------- |\n| Slash | `\u002Fwhathappened Kimi K3` |\n| Natural language | \"what are people on X saying about the OpenAI board drama\" |\n\nThe agent should auto-invoke when the description matches (launch reaction, X\u002FTwitter sentiment, \"what happened with…\").\n\n## Package layout\n\n```\nskills\u002Fwhathappened\u002F\n  SKILL.md                 # agent contract (pipeline + output template)\n  references\u002F\n    query-patterns.md      # X advanced search recipes\n    failure-modes.md       # thin sample, bots, entity collisions, …\n```\n\nShips as an [`npx skills`](https:\u002F\u002Fgithub.com\u002Fvercel-labs\u002Fskills) package (`skills\u002F\u003Cname>\u002FSKILL.md`) for install convenience. Runtime target remains Grok Build only.\n\n## Development\n\n```sh\n# edit the skill\n$EDITOR skills\u002Fwhathappened\u002FSKILL.md\n\n# smoke-test discovery locally\nnpx skills add .\u002F -l\n\n# install from this checkout for Grok\nnpx skills add .\u002F -g -y\n# or symlink into ~\u002F.grok\u002Fskills\u002Fwhathappened\n```\n","whathappened 是一个面向 X（原 Twitter）平台的智能体技能，专为实时事件感知与舆论分析设计。它基于 Grok Build 框架，自动聚合 X 上的原始帖文，生成结构化简报：包括事件核心事实、讨论热度分布、主流观点阵营划分、关键辩论焦点及原始帖链接。技术上采用 X 原生数据源优先、自适应时间窗口（分钟级至日级）、多维度采样（顶帖\u002F最新\u002F语义\u002F官方账号）与轻量 Web 辅助实体解析。适用于需要快速把握突发舆情、产品发布反馈或公共事件舆论动向的运营、公关、分析师及开发者场景。",2,"2026-07-18 02:30:07","CREATED_QUERY"]