Thanks to visit codestin.com
Credit goes to github.com

Skip to content
View ruodou233's full-sized avatar

Block or report ruodou233

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
ruodou233/README.md

错误乱码的 AI 工作流|Practical AI Agent Workflows

Open-source skills and workflows to help AI agents review their work, manage costs, coordinate long-running tasks, and retain reusable lessons.

我把日常与 AI Agent 协作中用到的任务编排、独立审查、成本优化和经验沉淀方法,做成可复用的开源 Skill,帮助 Agent 减少返工、接续长任务,并扩展研究与创作能力。

这些 Skill 面向 Claude Code / Codex 等支持 SKILL.md 的平台,可由 Agent 按你的电脑、订阅和已装工具调整接入方式。

这些 skill 遵循同一条分发原则:当多个 Agent 需要接入同一权威源时,采用单源维护——核心逻辑只改一处,减少同一规则在多处各自演化、重复漂移的风险;设计原则和两种实现结构(整包共享 / 共享规范+平台适配层)见 agent-orchestration 的"Skill 结构原则"一节。

可以拿来干什么

  • 提高交付质量,减少人工返工:用 cross-review 让另一家 AI 独立审查代码与方案,检查正确性、复杂度和替代思路;用 upgrade-audit 把对话中的偏好、踩坑经验沉淀进记忆与 Skill,减少重复纠正。
  • 算清成本,合理安排任务:用缓存保温先实测命中率和成本,确认划算再启用;配合额度监测,判断重任务何时运行、如何安排消耗。
  • 让 Agent 接续完成长任务:用 agent-orchestration 拆分任务、协调分工、衔接结果;用 connect-computers 接入远端电脑与 Agent,让多机资源参与工作。
  • 扩展 Agent 获取信息和操作网页的能力:用 web-access 处理检索、动态网页与需要登录态的页面;结合领域探索和社区口碑,为分析与决策补充可追溯的资料和真实使用反馈。
  • 把研究推进到可实施的方案:让聪明买手比较候选、成本和取舍,再用产品方案 Skill 明确首版范围、实现阶段与验收标准,形成可交给开发 Agent 的说明。
  • 把资料和想法做成作品:转写会议、访谈与播客,审阅中文里的套路表达;制作网页、演讲和作品集,让视觉、动效与声音共同服务内容。

开源 Skill 目录

以下各表按 GitHub Star 数降序排列,同星保留原顺序。排序核验:2026-09-13;徽章显示最新星标数。

Skill Stars 一句话介绍
de-ai-taste GitHub Stars AI 写的中文一眼就能看出来?逐条找出 AI 味、给出具体修改建议,让文章、讲稿和文案读起来像你写的。
Review AI-written Chinese and suggest edits to remove formulaic phrasing while preserving the author's voice.
cross-review GitHub Stars AI 的活总差一点,总要你擦屁股,总打丑补丁?让另一家 AI 挑刺复查,自己把活干完整,不用你一直兜底。
An agent skill for independent code and design reviews across AI providers, checking correctness, complexity, and better approaches.
claude-cache-keepalive GitHub Stars 缓存保温:实测命中、算清收益,让长会话少花冤枉 token
Measure prompt cache hits and costs, then configure automatic keepalive when the savings justify it.
free-token-eggs GitHub Stars 免费额度找半天、领一圈,模型还不好用?帮你挑值得领的免费 AI token——太蠢的大模型不收录,吃点好的。
Find worthwhile free LLM API credits and AI trials, with eligibility, expiry dates, and claim links checked.
domain-explorer GitHub Stars 速通新领域:入门、转行、选课题,先把来龙去脉和各路说法弄明白
Get up to speed on a new topic through its history, competing approaches, expert debates, and practical experience.
upgrade-audit GitHub Stars 把你教过 AI 的东西留下来:沉淀偏好、复盘踩坑、更新 skill 和工作流程
Review conversation history, agent memory, and skills to identify reusable lessons and propose updates to outdated instructions.
smart-buyer GitHub Stars AI 推荐了一堆,到底该选哪个还是没弄明白?聪明买手帮你把商品、软件、服务和技术方案研究明白再选。
Research and compare products, software, services, and technical options using buying criteria, verified costs, and real user feedback.
agent-orchestration GitHub Stars 复杂任务跑到半夜,你不可能一直盯着。让 Agent 分工跑长任务和批量工作,你只管第二天早上收结果。
Coordinate AI agents for long-running tasks, parallel work, and overnight workflows.
audio-transcribe GitHub Stars 把会议、访谈、课程和播客变成文字稿,需要说话人、时间戳和字幕也能安排
Transcribe Mandarin recordings, podcasts, meetings, and videos with speaker labels, timestamps, subtitles, and transcript review.
community-buzz GitHub Stars 想听真实口碑?从社区里挖使用体验、竞品痛点和技术争议
Find real user experiences, product comparisons, and disagreements in online communities, with links to the original discussions.
craft-frontend-ppt GitHub Stars 让网页成为舞台:演讲、作品集、品牌页面和交互文章,让视觉、动效与声音一起表达
Create HTML slide decks, portfolios, landing pages, and interactive stories with visual design, animation, and sound.
wisdom-roundtable GitHub Stars 大事拿不定,小事没思路?拉一桌不同视角,看看盲区和分歧
Explore ideas and decisions through multiple mental models, then compare their assumptions, blind spots, agreements, and disagreements.
improve-product-plan GitHub Stars 想做成一个作品?把应用、工具、自动化和游戏想法打磨成能开工的方案
Plan product requirements and MVP scope, then produce a development spec with milestones and acceptance criteria.
connect-computers GitHub Stars 把电脑连起来:轻薄本调家中工作机,闲置电脑跑任务,出门也能接着干
Set up SSH, VPN, remote desktop access, and remote AI agents across your computers for work at home and on the go.
codex-reset-watch GitHub Stars Codex 额度哨兵:看看还能蹬多久,重任务该现在上还是缓一缓
Track Codex usage limits, spending pace, and reset signals to plan your next tasks.
web-access GitHub Stars 给 AI 装上联网能力:查资料、读动态网页、用登录态,把网页上的事办完
An agent skill for web research and browser automation, including logged-in pages, dynamic content, and browser history search.

机制参考仓

这些仓库把我的具体实现放出来,按各仓说明接到你自己的环境。

Skill Stars 一句话介绍
claude-code-feishu GitHub Stars 出门前电脑上开的活,到手机上就接不上了?用飞书接着指挥同一个 Claude Code 会话,回来接着干。
Continue the same desktop Claude Code session from your phone through Feishu / Lark.
growth-radar GitHub Stars 想学爆款先看清涨粉从哪来:筛案例、找作品、拆内容
Research social media follower growth through creator rankings, published content, and public comments.
polymarket-anomaly-watch GitHub Stars 每天醒来就有一份科技情报——市场异动、GitHub 热榜、App Store 榜单,AI 替你盯着。
Monitor Polymarket anomalies, GitHub trends, and App Store rankings for a daily tech briefing.
multi-upstream-agent-routing GitHub Stars 给 Agent 挑线路:核能力、算实付、配主备,把上游选择理清楚
A design guide for multi-provider LLM routing, fallback policies, capability checks, and cost accounting.
turn-guard GitHub Stars Claude Code 老忘流程?把该做的检查变成每回合提醒
Claude Code hooks for background task follow-ups and reminders to review plans and completed changes.

找到我

  • 小红书:错误乱码
  • 微信公众号:能工智人错误乱码
  • B站:若逗道人

有修改方案或问题,欢迎在对应仓库提 issue / PR。

Popular repositories Loading

  1. de-ai-taste de-ai-taste Public

    AI 写的中文一眼就能看出来?逐条找出 AI 味、给出具体修改建议,让文章、讲稿和文案读起来像你写的。 Review AI-written Chinese and suggest edits to remove formulaic phrasing while preserving the author's voice.

    21

  2. cross-review cross-review Public

    AI 的活总差一点,总要你擦屁股,总打丑补丁?让另一家 AI 挑刺复查,自己把活干完整,不用你一直兜底。 An agent skill for independent code and design reviews across AI providers, checking correctness, complexity, and better approaches.

    15

  3. claude-cache-keepalive claude-cache-keepalive Public

    缓存命中可以节省 90% token。如果你五分钟内看不完 AI 的报告,可以考虑缓存保温。

    13

  4. free-token-eggs free-token-eggs Public

    免费额度找半天、领一圈,模型还不好用?帮你挑值得领的免费 AI token——太蠢的大模型不收录,吃点好的。 Find worthwhile free LLM API credits and AI trials, with eligibility, expiry dates, and claim links checked.

    Python 9 3

  5. ruodou233 ruodou233 Public

    错误乱码的 AI Agent 工作流:研究、选择、写作、做作品,让 AI 把活干完整。 Open-source agent skills and workflows for research, decision making, writing, and creative work.

    7

  6. agentops-skills agentops-skills Public

    作者本机实测的 AI Agent skill 全家桶:调研选型、写作创作、长任务协作,一套装好。 A collection of practical agent skills for research, writing, product planning, and task orchestration.

    Shell 6 1