CF lobster-publish-local
帮你在本机一键发布到小红书、知乎、视频号、公众号、抖音、快手、微博、西瓜视频、B站、百家号、网易号、搜狐号等"用扫码登录"的平台，并支持短视频帮发（11 平台全量打通，含流量机制与 AI 可引用规范）。解决云端 RPA 被风控封号的问题——你的账号在你电脑登录，发布也在你电脑，零风控。 适配 WorkBuddy 桌面端：扫码 cookie 绑定 + 浏览器抓取 + 平台发布执行 + 结果回写龙虾 GEO 发布中心（数据闭环）。 v2.22.2 防白嫖门禁升级：彻底删除 30 天免费体验硬编码兜底；严格按「邀请码 + API Key 双凭证（同账号）」识别会员身份与权益有效期。 v2.22.5 客户本地图片上传（大当家铁令）：客户自己配图 = 真实图片文件，严禁只存 URL 字符串。新增 MCP `geo_upload_hook_image` + REST `POST /v1/user/hook/upload`，服务端落盘并返回公网 URL。 v2.23.0 短视频分发（只分发不创作）：新增 MCP `geo_short_video_spec` / `geo_short_video_publish` / `geo_short_video_result`；支持视频号/抖音/快手/小红书/微博 5 平台，复用 `geo_bind_cookie` 跳起浏览器扫码统一仪式，本地视频文件不上云。 v2.24.0 规范入口统一 + 流程重写（大当家拍板）：新增 MCP `geo_content_form_select` / `geo_content_spec`（发布意图先反问内容形态：长文/短图文/短视频/问答 → 再给该形态在目标平台的硬规范）+ `geo_lobster_news_stats`（龙虾资讯官网方案B完整数据闭环：阅读量/独立阅读量/AI引用率/获客，防遗忘）。 v2.24.1 官网发布前铁律 + 公众号级排版（2026-08-27 大当家补刀）：龙虾资讯官网发布前顺序锁死「①内容形态反问 ②引流钩子问询 ③HTML 公众号级排版 ④预览确认」四步，客户不说"发吧"绝不发布；排版复刻公众号（首字下沉/首行缩进/加粗/居中/衬线字体/配图带图注），分类后端自动识别不追问。 v2.24.2 排版铁律全平台通用化（2026-08-27 大当家钉死）：HTML 是所有平台永远的最高优先级输出格式，markdown 仅作备选且必须富排版，禁初级裸文本/文字扎堆/乱码；发布前排版自检硬性闸门。 v2.25.0 配图三级优先级 + 全局横版 + 封面带字（2026-08-27 收官）：配图逻辑锁死「①平台原生配图能力优先(知乎/头条/百家等直接蹭) ②无原生能力且客户选AI配图→WorkBuddy自带生图(B方案:底图+PIL叠字) ③客户自配图→真实文件上传(geo_upload_hook_image)」；正文配图一律横版(16:9)禁正方形；封面图可带主题文案(统一B方案叠字)；生图非必要不乱烧钱。 v2.25.1 公众号授权入口域名统一（2026-08-29 大当家紧急修复）：后端 `/v1/wechat-thirdparty/jump` 改为必须用户主动点击「前往微信授权」按钮后才跳转 `mp.weixin.qq.com`，保证 referer 必定来自 `geo.aiskillhub.vip`，彻底修复"授权入口页所在域名：空"。Skill 侧统一以 `geo.aiskillhub.vip` 为公众号 OAuth 唯一入口域名，禁止直接构造或打开微信授权 URL。 v2.25.2 WorkBuddy 公众号 OAuth 全闭环（2026-08-29 大当家核验收尾）：本地执行器 `auto-bind.js` 新增 `wechat_official` OAuth 模式，在 WorkBuddy 中调用 GEO内容发布助手发公众号时——① 调 `geo_bind_platform(wechat_official)` 取 `auth_url`；② 本地 agent 自动弹起系统浏览器打开中转页并点击「前往微信授权」；③ 客户用管理员微信扫码确认；④ 授权成功后浏览器自动关闭；⑤ 智能识别运行环境（WorkBuddy 桌面端/官网网页端/纯 MCP 客户端）并引导客户下一步（直接发布 or 先创作内容）。 v2.25.3 前端授权弹窗全国统一（2026-08-29 大当家钉死）：官网网页端所有平台授权绑定弹窗（公众号、Cookie 平台、龙虾资讯官网、Connector 市场 OAuth）彻底删除手绘/静态二维码图片，统一为「跳起浏览器扫码/授权」按钮；绑定成功后按当前页面环境智能引导下一步（携带文章则直接打开发布弹窗，否则提示去发布中心
帮你在本机一键发布到小红书、知乎、视频号、公众号、抖音、快手、微博、西瓜视频、B站、百家号、网易号、搜狐号等"用扫码登录"的平台，并支持短视频帮发（11 平台全量打通，含流量机制与 AI 可引用规范）。解决云端 RPA 被风控封号的问题——你的账号在你电脑登录，发布也在你电脑，零风控。 适配 WorkBuddy…
As a process F 34/100 · Will not run — References files that are not bundled: CHANGELOG.md, assets/pet-greeting.html, scripts/ensure-engine.js
How to improve
- Say in the description WHEN to use the skill ("use when…", example requests): that is the agent's main cue.
- Shorten the description to 1024 characters.
- For Hermes the description must be one sentence under 60 characters; move the conditions to a "When to Use" section.
- The SKILL.md body is over 5,000 tokens: move reference detail into references/ and load it when needed.
- The text references files that are not there: add them or drop the references.
- Your own cases (evals/evals.json, 4–6 real requests with expected answers): the full check would then run those instead of a model-drafted suite.
- A spec.yaml with trigger phrases and assertions — a behaviour contract for CI; `skilltest init` writes a template.
Guard findings · 0
✓ No critical or high findings
Files scanned: 2. Evidence is masked. Grey chips explain why severity was lowered.
Against the Agent Skills spec
- error
description-longdescription is 3724 chars, limit 1024 - warning
description-long-hermesdescription is 3724 chars; the Hermes authoring standard requires ≤ 60 (one sentence, ending with a period) - warning
description-no-whenneither description nor a "## When to Use" section says when to use the skill - warning
body-longSKILL.md body ≈ 12158 tokens (recommended < 5000); move details to references/ - warning
missing-refreference to a missing file: CHANGELOG.md - warning
missing-refreference to a missing file: assets/pet-greeting.html - warning
missing-refreference to a missing file: scripts/ensure-engine.js - note
frontmatter-keyunknown frontmatter key "slug" - note
frontmatter-keyunknown frontmatter key "displayName" - note
frontmatter-keyunknown frontmatter key "language"
Process rating: all ten parameters 34/100
- 0Tools and files. 3 referenced file(s) missing: CHANGELOG.md, assets/pet-greeting.html, scripts/ensure-engine.js
- 0Result and completion. Does not say what the result is
- 0Inputs and preconditions. Does not say what the process needs to start
- 0Progress reporting. Says nothing while it works
- 20When it triggers. No condition that starts the skill
- 30Running it twice. 19 mutating operations with no state check
- 40Execution cost. Instruction body is 12158 tokens: crowds the task out of the window
- 50Failures and branches. 0 branches, has a failure section
- 100Steps. 296 steps
- 100Consistency. Name and required fields are in place
- low 29 top-level sections: this looks like several domains in one skill
- low The response is described with custom markup (7 tags): a typed call is more reliable
Everything here is measured from the skill text rather than judged by a model, so the numbers are checkable. A parameter weighs more when it is a more common reason for the process to stall.
Quality signals
- +4Description does not say when NOT to use the skill (false activations)
- +3Description length 3723: 120–800 characters recommended
- +3Output format is not stated: the model decides each time
- -293 emoji in the instructions: noise for the model
- +2Single-language instructions
- +5Description quotes 7 example trigger phrases
- +4Structure: 89 headings
- +3Step-by-step instructions: 296 items
- +4Has examples (9 code blocks)
- +1License stated
Quality base 70; lint remarks subtract, signals add up to 100. Result: 17.