# Simon's Skill Directory — complete bilingual text Simon Lin's free personal sharing site. 网站：https://agentskill.nz/ · English：https://agentskill.nz/en/ Structured catalog: https://agentskill.nz/catalog.json Read the small index first when possible: https://agentskill.nz/llms.txt ## Collection status / 模块状态 - [Open-source skills and tools](https://agentskill.nz/open-source/) · [English](https://agentskill.nz/en/open-source/) - [Personal skills — reserved](https://agentskill.nz/personal/) · [English](https://agentskill.nz/en/personal/) - [GitHub recommendations — reserved](https://agentskill.nz/recommendations/) · [English](https://agentskill.nz/en/recommendations/) 1. Open-source skills and tools / 已开源 Skill 与工具: 12 published entries, including 5 skills and 7 other tools. Source baseline checked 2026-10-21 (Asia/Singapore). 2. Simon personal skills / Simon 个人 Skill: reserved, 0 entries. 暂未开放，等待 Simon 清单。 3. GitHub public recommendations / GitHub 公开 Skill 推荐: reserved, 0 entries. 清单稍后补充。 No unpublished personal skills or third-party recommendations are included. Downloads are public fixed-revision source archives, not installers. Financial projects provide data and research tools only, not investment advice. --- ## About Simon 林 / 个人简介 - Profile source: Simon's self-description and supplied avatar - Supplied on: 2026-10-22 - Image: https://agentskill.nz/images/simon-profile-2026-10-22.jpeg - Contact: simonlin0423@gmail.com - Douyin / 抖音: Simon林 - WeChat Channels / 视频号: Simon.林 这里是我的个人免费 Skill 分享站。 践行 Vibe Coding 和 OPC 的独立开发者 不会写代码 · 金融行业出身 · Top2 硕士 AI 业务咨询与落地：simonlin0423@gmail.com My personal sharing site for free skills and tools. Independent developer practicing Vibe Coding and OPC No coding background · Finance professional · Master’s graduate from a top-two university AI business consulting and implementation: simonlin0423@gmail.com This profile is separate from the reserved personal-skill collection. It does not add any skill entries. --- # a-stock-data - Name: a-stock-data - Collection: Simon open-source skills and tools - Type: Skill / data tool - Author: Simon Lin - Source: https://github.com/simonlin1212/a-stock-data - README: https://github.com/simonlin1212/a-stock-data/blob/29b07d2e37005b9fd568bca1262f6e572b5f9407/README.md - Source commit: 29b07d2e37005b9fd568bca1262f6e572b5f9407 - License: Apache-2.0 - Version: v3.10.2 - Formal release: yes - Release date: 2026-10-11 - Source verified: 2026-10-21 (Asia/Singapore) - Download: https://agentskill.nz/downloads/a-stock-data-v3.10.2.zip - Bytes: 418455 - SHA-256: 82673d1131569adfe2e242a87e19dd879583c0ccc826381121b41c18e87850ea - Chinese page: https://agentskill.nz/projects/a-stock-data/ - English page: https://agentskill.nz/en/projects/a-stock-data/ ## 中文 给 AI 编程助手读取 A 股及相关市场原始数据的 Skill。 - 覆盖行情、K 线、财务、公告、研报与资金数据。 - 提供宏观、期货、期权与可转债等数据入口。 - 在 Skill 中说明数据口径、来源与备用路线。 支持与接入：Claude Code、Codex、OpenClaw、WorkBuddy、Muse、Grok Bot；本质是结构化 Markdown 与内嵌 Python。 1. 下载并解压 ZIP，保留 SKILL.md 和随附许可证。 2. Claude Code 用户把 SKILL.md 放进自己的 skills 目录；其他 Agent 可将其加入项目上下文。 3. 安装 README 列出的 Python 依赖，让 Agent 按 Skill 中的取数示例使用。 ```text pip install mootdx requests pandas stockstats numpy baostock xlrd openpyxl ``` 这是取数代码与使用说明，不是数据订阅。部分来源需要用户自己的鉴权，使用时遵守各数据源条款。 ## English A skill for AI coding assistants to retrieve raw A-share and related market data. - Covers quotes, K-lines, financials, filings, research reports and fund flows. - Includes macro, futures, options and convertible-bond data routes. - Documents source conventions and fallback routes inside the skill. Agents and compatibility: Claude Code, Codex, OpenClaw, WorkBuddy, Muse and Grok Bot. The skill is structured Markdown with embedded Python. 1. Download and extract the ZIP, keeping SKILL.md and its license. 2. Place SKILL.md in your Claude Code skills directory, or add it to another agent's project context. 3. Install the Python dependencies listed in the README and use the retrieval examples. ```text pip install mootdx requests pandas stockstats numpy baostock xlrd openpyxl ``` The archive distributes retrieval code and instructions, not a data subscription. Some sources require your own authentication and have separate terms. Only data and research tools. Not investment advice. 只提供数据和研究工具，不构成投资建议。 Source and archive checks do not establish complete runtime or feature verification. --- # Agent-Staff - Name: Agent-Staff - Collection: Simon open-source skills and tools - Type: Workbench / framework - Author: Simon Lin - Source: https://github.com/simonlin1212/Agent-Staff - README: https://github.com/simonlin1212/Agent-Staff/blob/edc782b2f8c5d037d24bd63067728f549dd76f53/README.md - Source commit: edc782b2f8c5d037d24bd63067728f549dd76f53 - License: MIT - Version: 2026-10-21 - Formal release: no; default-branch source snapshot - Release date: none - Source verified: 2026-10-21 (Asia/Singapore) - Download: https://agentskill.nz/downloads/Agent-Staff-2026-10-21.zip - Bytes: 1600197 - SHA-256: 2210e7fda6456ad6a5f62f8b7bf6dea2aa432b775e9e268fc74a3f9bae9031b0 - Chinese page: https://agentskill.nz/projects/agent-staff/ - English page: https://agentskill.nz/en/projects/agent-staff/ ## 中文 按组织架构配置部门 Agent，在飞书里协作的公司工作框架。 - 为业务、财务、人力等部门配置不同 Agent。 - 通过参谋长角色汇总跨部门信息。 - 用飞书多维表格保存结构化记录，并设置部门访问边界。 支持与接入：作者实测 Claude；也提供 DeepSeek、Minimax、Qwen、GLM、OpenRouter 与 Ollama 等接口配置，未逐一实测。 1. 下载并解压，先阅读部署与飞书接入指南。 2. 准备自己控制的飞书应用，运行安装和配置向导。 3. 填入自己的应用凭据与模型配置后启动。 ```text bash install.sh bash setup.sh bash 启动.sh ``` 安装脚本会安装外部依赖；先检查内容与系统要求。使用者自备飞书应用与模型服务。 ## English A company-agent framework that mirrors departments and collaborates inside Feishu. - Assigns separate agents to business, finance, HR and other departments. - Uses a chief-of-staff role to aggregate cross-department information. - Keeps structured records in Feishu tables with department access boundaries. Agents and compatibility: The author tested Claude. Configuration also covers DeepSeek, Minimax, Qwen, GLM, OpenRouter and Ollama, which have not each been tested. 1. Extract the source and read the deployment and Feishu connection guides. 2. Prepare Feishu applications you control, then run the installer and setup wizard. 3. Enter your own application credentials and model configuration, then start the service. ```text bash install.sh bash setup.sh bash 启动.sh ``` The installer adds external dependencies; review it and its system requirements first. Bring your own Feishu applications and model service. Source and archive checks do not establish complete runtime or feature verification. --- # FactReach - Name: FactReach - Collection: Simon open-source skills and tools - Type: Skill / data tool - Author: Simon Lin - Source: https://github.com/simonlin1212/FactReach - README: https://github.com/simonlin1212/FactReach/blob/2b358e1792419f3346b0606531962a424601fa9a/README.md - Source commit: 2b358e1792419f3346b0606531962a424601fa9a - License: MIT - Version: v1.0.0 - Formal release: yes - Release date: 2026-09-02 - Source verified: 2026-10-21 (Asia/Singapore) - Download: https://agentskill.nz/downloads/FactReach-v1.0.0.zip - Bytes: 768459 - SHA-256: 8b541e1470b84ab07bcf50f651ec3d3e795c734ea39a5d8d910a596f81f725e6 - Chinese page: https://agentskill.nz/projects/factreach/ - English page: https://agentskill.nz/en/projects/factreach/ ## 中文 让 Agent 先搜索、再回答的互联网搜索与网页读取工具。 - 为公开网页、社交平台、视频和知识源选择读取路线。 - 用 doctor 检查各渠道状态，显示后端与修复提示。 - 可检查和预览 Agent 的搜索规则；写入规则需要明确授权。 支持与接入：Claude Code、Codex，以及支持 Shell 的 AI Agent。渠道可用性取决于本机依赖、登录态与自备凭据。 1. 下载并解压本站源码 ZIP，在独立 Python 环境中安装；已有 Agent-Reach 的环境须先按 README 迁移。 2. 运行 factreach doctor 检查渠道；默认 install 仅检查环境。 3. 使用 factreach web read 读取公开网页，再按安装指南启用需要的可选后端。 ```text python3 -m venv .venv source .venv/bin/activate pip install . factreach doctor ``` 基于 Agent-Reach 延续开发，原版权、LICENSE 与 NOTICE 保留在下载包中。需要外部服务的渠道可能有自己的费用或限制。 ## English Internet search and page reading that help agents check evidence before answering. - Routes public pages, social platforms, video and knowledge sources to suitable readers. - Checks channel availability and reports backends and repair hints with doctor. - Previews and checks agent search policies; writing a policy requires explicit permission. Agents and compatibility: Claude Code, Codex and AI agents with shell access. Availability depends on local tools, sessions and user-supplied credentials. 1. Download and extract the source ZIP, then install in an isolated Python environment. Existing Agent-Reach users should follow the README migration steps. 2. Run factreach doctor to check channels. The default install command only checks the environment. 3. Read a public page with factreach web read, then follow the guide to enable optional backends. ```text python3 -m venv .venv source .venv/bin/activate pip install . factreach doctor ``` Continues the Agent-Reach project. Original copyright, LICENSE and NOTICE remain in the archive. Optional external services may have their own costs or limits. Source and archive checks do not establish complete runtime or feature verification. --- # findadoctor-mcp - Name: findadoctor-mcp - Collection: Simon open-source skills and tools - Type: MCP / experiment - Author: Simon Lin - Source: https://github.com/simonlin1212/findadoctor-mcp - README: https://github.com/simonlin1212/findadoctor-mcp/blob/e740b49368f2c2e3c23642d8424414507f790ae3/README.md - Source commit: e740b49368f2c2e3c23642d8424414507f790ae3 - License: MIT - Version: 2026-10-21 - Formal release: no; default-branch source snapshot - Release date: none - Source verified: 2026-10-21 (Asia/Singapore) - Download: https://agentskill.nz/downloads/findadoctor-mcp-2026-10-21.zip - Bytes: 794769 - SHA-256: 544edb7665e373f2e9cbddc8eeebc7182e5e43bb203c6fe7835220d68b3ce4a3 - Chinese page: https://agentskill.nz/projects/findadoctor-mcp/ - English page: https://agentskill.nz/en/projects/findadoctor-mcp/ ## 中文 让 AI Agent 按需求和地点查找医生，并核对公开来源的 MCP 接入资料。 - 提供 FindADoctor.si 远程 MCP 的连接指南。 - 按就医国家、地点、专科或需求检索医生资料。 - 附带工具参考、示例与 findadoctor Agent Skill。 支持与接入：Claude、ChatGPT、Claude Code、Codex CLI、Gemini CLI、Cursor、VS Code；远程 MCP 能力以客户端和套餐为准。 1. 下载 ZIP，阅读连接指南与工具说明；此包是指南和 Skill，不是医生数据库。 2. 将远程 MCP 地址加入支持 Streamable HTTP 的客户端。 3. 先确定就医国家，再查询和核对来源；保险和接诊情况需另外确认。 ```text codex mcp add findadoctor --url https://api.findadoctor.si/mcp ``` 不构成医疗建议。数据库覆盖、额度与服务状态以 FindADoctor.si 当前说明为准；MIT 许可适用于此仓库。 ## English MCP connection guides that help AI agents find doctors and check public-source evidence. - Documents the FindADoctor.si remote MCP connection. - Supports doctor searches by country, location, specialty or care need. - Includes tool references, examples and the findadoctor Agent Skill. Agents and compatibility: Claude, ChatGPT, Claude Code, Codex CLI, Gemini CLI, Cursor and VS Code. Remote MCP support depends on the client and plan. 1. Download the guides and skill. This archive is not the doctor database. 2. Add the remote MCP address to a client that supports Streamable HTTP. 3. Establish the care country first, then search and check sources. Confirm insurance and appointment availability separately. ```text codex mcp add findadoctor --url https://api.findadoctor.si/mcp ``` Not medical advice. Current coverage, allowances and service availability are described by FindADoctor.si. The MIT license applies to this repository. Source and archive checks do not establish complete runtime or feature verification. --- # finwisp-home - Name: finwisp-home - Collection: Simon open-source skills and tools - Type: MCP / experiment - Author: Simon Lin - Source: https://github.com/simonlin1212/finwisp-home - README: https://github.com/simonlin1212/finwisp-home/blob/59714719abc728930a2b8f58e4e052db4afd0972/README.md - Source commit: 59714719abc728930a2b8f58e4e052db4afd0972 - License: Apache-2.0 - Version: 2026-10-21 - Formal release: no; default-branch source snapshot - Release date: none - Source verified: 2026-10-21 (Asia/Singapore) - Download: https://agentskill.nz/downloads/finwisp-home-2026-10-21.zip - Bytes: 1476781 - SHA-256: 86158a532dc39deed9715c819aab4bb8bc43c44c1a91e3ddb910870584d00902 - Chinese page: https://agentskill.nz/projects/finwisp-home/ - English page: https://agentskill.nz/en/projects/finwisp-home/ ## 中文 基于 SEC 财报、学习卡片与小测验的家庭金融学习作品。 - 读取 SEC EDGAR 披露并计算财务指标。 - 提供概念解释、学习卡片、测验与本地家庭学习记录。 - 支持文字输入和依赖浏览器能力的语音交互。 支持与接入：Streamable HTTP MCP；README 记录了 Codex CLI 验证。Claude Code 配置示例未由作者实测；Codex 语音仍为实验预览。 1. 下载解压，准备 README 要求的 Node 版本并安装依赖。 2. 通过环境变量提供自己的 SEC 联系邮箱。 3. 运行 Home Hub，打开本地 8796 端口；MCP 与语音设置见 README。 ```text npm ci # Set SEC_CONTACT_EMAIL to your own contact npm run hub ``` 教育用途。浏览器语音可能由浏览器供应商远程处理；可始终使用文字。该作品的比赛与工具使用状态以 README 说明为准。 ## English A family finance-learning project built around SEC filings, lesson cards and quizzes. - Reads SEC EDGAR disclosures and computes financial metrics. - Provides concept explanations, learning cards, quizzes and local household learning records. - Accepts text and browser-dependent voice interaction. Agents and compatibility: Streamable HTTP MCP. The README records Codex CLI verification. The Claude Code configuration was not tested by the author; Codex voice remains experimental. 1. Extract the source, prepare the documented Node version and install dependencies. 2. Set your own SEC contact email through the environment. 3. Run the Home Hub and open local port 8796. Read the README for MCP and voice setup. ```text npm ci # Set SEC_CONTACT_EMAIL to your own contact npm run hub ``` Educational use. Browser voice services may process audio remotely; text remains available. See the README for the project's competition and tooling status. Only data and research tools. Not investment advice. 只提供数据和研究工具，不构成投资建议。 Source and archive checks do not establish complete runtime or feature verification. --- # global-stock-data - Name: global-stock-data - Collection: Simon open-source skills and tools - Type: Skill / data tool - Author: Simon Lin - Source: https://github.com/simonlin1212/global-stock-data - README: https://github.com/simonlin1212/global-stock-data/blob/cc6afb5bad45715a7395fcccb09d786d8a1c160d/README.md - Source commit: cc6afb5bad45715a7395fcccb09d786d8a1c160d - License: Apache-2.0 - Version: v2.2.0 - Formal release: yes - Release date: 2026-10-12 - Source verified: 2026-10-21 (Asia/Singapore) - Download: https://agentskill.nz/downloads/global-stock-data-v2.2.0.zip - Bytes: 241451 - SHA-256: 80bef16f1a30a68299a51ed7da30363e17690a6d819da7c825806f55db5dd972 - Chinese page: https://agentskill.nz/projects/global-stock-data/ - English page: https://agentskill.nz/en/projects/global-stock-data/ ## 中文 官方来源优先的美股与全球市场数据 Skill。 - 读取行情、公司基本面、资金与 SEC 披露。 - 提供期权、波动率、宏观利率与外汇等数据路线。 - 按来源标注使用边界与合规层级。 支持与接入：Claude Code、Codex、OpenClaw，以及能读取 Markdown 上下文的 AI 编程助手。 1. 下载并解压源码，找到 SKILL.md。 2. 将 Skill 放入 Agent 的技能目录或项目上下文。 3. 安装 requests；Windows 用户按 README 加装 tzdata，再运行 Skill 中的例子。 ```text pip install requests # Windows pip install tzdata ``` 分发的是代码和说明；数据可用性与使用条件由上游来源决定。 ## English A US and global market data skill that prioritizes official sources. - Retrieves quotes, company fundamentals, flows and SEC disclosures. - Includes options, volatility, macro rates and foreign-exchange routes. - Labels source-specific usage boundaries and compliance tiers. Agents and compatibility: Claude Code, Codex, OpenClaw and AI coding assistants that accept Markdown context. 1. Download and extract the source, then locate SKILL.md. 2. Add the skill to your agent's skills directory or project context. 3. Install requests; Windows users should also install tzdata as described in the README, then use the included examples. ```text pip install requests # Windows pip install tzdata ``` This is a distribution of code and documentation. Upstream sources determine data availability and usage conditions. Only data and research tools. Not investment advice. 只提供数据和研究工具，不构成投资建议。 Source and archive checks do not establish complete runtime or feature verification. --- # Hiring-Radar - Name: Hiring-Radar - Collection: Simon open-source skills and tools - Type: Skill / data tool - Author: Simon Lin - Source: https://github.com/simonlin1212/Hiring-Radar - README: https://github.com/simonlin1212/Hiring-Radar/blob/dae32964f9c3b7289a85eac8693225e5248e7c82/README.md - Source commit: dae32964f9c3b7289a85eac8693225e5248e7c82 - License: MIT - Version: v2.0.0 - Formal release: yes - Release date: 2026-10-12 - Source verified: 2026-10-21 (Asia/Singapore) - Download: https://agentskill.nz/downloads/Hiring-Radar-v2.0.0.zip - Bytes: 305530 - SHA-256: 2bc8e89e9a26bf8f4a7a948d23f4f97991607252326af1088cf5f807318edc8b - Chinese page: https://agentskill.nz/projects/hiring-radar/ - English page: https://agentskill.nz/en/projects/hiring-radar/ ## 中文 读取企业公开招聘信息，用于求职与产业研究的本地工具。 - 按公司、国家与关键词查找在招岗位。 - 将多来源职位整理成统一字段，并支持去重与续跑。 - 用本地存档对照新增、下架与历史变化。 支持与接入：Claude Code、Codex 等可通过 MCP 调用；也可直接运行 Python 命令。 1. 下载解压，准备 Python 3。 2. 列出支持来源，按需下载公开公司名录。 3. 按公司查询，或按 README 配置 --mcp 供 Agent 调用。 ```text python3 hiring_radar.py --list python3 hiring_radar.py anthropic --keyword research ``` 只读取主动公开的岗位，不抓个人简历或需要登录的内容。招聘变化可作研究线索，不能单独证明公司经营结果。 ## English A local reader for public job postings, for job searches and industry research. - Finds open roles by company, country and keyword. - Normalizes postings across sources with deduplication and resumable runs. - Stores local snapshots to compare new, removed and historical roles. Agents and compatibility: Claude Code and Codex can call it over MCP. It also works as a Python command-line tool. 1. Download and extract the source, and prepare Python 3. 2. List sources and optionally download the public company directory. 3. Query a company, or follow the README to expose --mcp to your agent. ```text python3 hiring_radar.py --list python3 hiring_radar.py anthropic --keyword research ``` Reads actively published roles, not private resumes or login-only content. Hiring changes are research leads and do not establish business outcomes on their own. Only data and research tools. Not investment advice. 只提供数据和研究工具，不构成投资建议。 Source and archive checks do not establish complete runtime or feature verification. --- # investment-news - Name: investment-news - Collection: Simon open-source skills and tools - Type: Skill / data tool - Author: Simon Lin - Source: https://github.com/simonlin1212/investment-news - README: https://github.com/simonlin1212/investment-news/blob/b94085f2f6fce82d233b1771580d96f66f89b595/README.md - Source commit: b94085f2f6fce82d233b1771580d96f66f89b595 - License: MIT - Version: v1.0.3 - Formal release: yes - Release date: 2026-08-20 - Source verified: 2026-10-21 (Asia/Singapore) - Download: https://agentskill.nz/downloads/investment-news-v1.0.3.zip - Bytes: 591910 - SHA-256: a232dd274c0d332d07fb41ea667ae540581b867864a157ecc0d6a8537f6d889f - Chinese page: https://agentskill.nz/projects/investment-news/ - English page: https://agentskill.nz/en/projects/investment-news/ ## 中文 在本地汇集全球产业资讯，用自己的 AI 整理中文要点。 - 按产业赛道查看公开信息源的新闻。 - 生成中文要点与翻译，并保留原文链接。 - 在本地浏览器看板触发抓取与刷新。 支持与接入：本机 Claude Code 订阅，或兼容 OpenAI 接口的模型 API；它本身是 Python 本地看板。 1. 下载解压，准备 Python 3.7 或以上版本。 2. 按 README 配置已登录的 Claude Code，或自己的模型 API。 3. 启动服务并打开本地 8793 端口，在页面点击刷新。 ```text python3 server.py ``` Python 部分使用标准库。AI 订阅和 API 由用户自备，本站不提供模型服务。 ## English A local industry-news dashboard that uses your own AI to prepare Chinese briefs. - Groups public news sources by industry sector. - Produces Chinese summaries and translations with links to the originals. - Runs collection and refresh from a local browser dashboard. Agents and compatibility: A local Claude Code subscription or an OpenAI-compatible model API. The project itself is a local Python dashboard. 1. Extract the ZIP and prepare Python 3.7 or newer. 2. Configure your signed-in Claude Code installation or your own model API following the README. 3. Start the server, open local port 8793 and use the refresh button. ```text python3 server.py ``` The Python code uses the standard library. Bring your own AI subscription or API; this site does not provide model services. Only data and research tools. Not investment advice. 只提供数据和研究工具，不构成投资建议。 Source and archive checks do not establish complete runtime or feature verification. --- # sovereign-market-brief - Name: sovereign-market-brief - Collection: Simon open-source skills and tools - Type: MCP / experiment - Author: Simon Lin - Source: https://github.com/simonlin1212/sovereign-market-brief - README: https://github.com/simonlin1212/sovereign-market-brief/blob/c0c6349436c3df8fc089997a32eef8d7ccbe68d4/README.md - Source commit: c0c6349436c3df8fc089997a32eef8d7ccbe68d4 - License: Apache-2.0 - Version: v1.0 - Formal release: yes - Release date: 2026-10-15 - Source verified: 2026-10-21 (Asia/Singapore) - Download: https://agentskill.nz/downloads/sovereign-market-brief-v1.0.zip - Bytes: 257343 - SHA-256: b5e88ae31400302157db58d0236e01d9fc8c32df2e92e767a335642124706411 - Chinese page: https://agentskill.nz/projects/sovereign-market-brief/ - English page: https://agentskill.nz/en/projects/sovereign-market-brief/ ## 中文 用确定性事实校验约束 Apertus 多语言市场简报的比赛作品。 - 先用 Python 计算本地 CSV 中的指标与事实编号。 - 让 Apertus 基于事实生成多语言简报。 - 校验数字、日期、方向与引用，未通过的草稿会被阻止发布。 支持与接入：使用 Apertus 与兼容 OpenAI 的模型端点；这是独立比赛作品，不是通用 MCP 服务。 1. 下载解压，阅读根目录和 track_2b 说明。 2. Docker 路线执行 make run，首次需另下载约 5 GB 模型并准备相应内存。 3. 也可按 README 配置已有 Apertus 端点，用 Python 处理自己的 date/value CSV。 ```text make run ``` 代码为 Apache-2.0，文档为 CC-BY-4.0。源码 ZIP 不包含另行下载的模型；模型和运行环境须自备。 ## English A competition project that verifies facts in multilingual Apertus market briefs. - Computes numbered facts and metrics from a local CSV with Python. - Asks Apertus to write multilingual briefs grounded in those facts. - Checks numbers, dates, direction and citations, blocking drafts that fail verification. Agents and compatibility: Apertus through an OpenAI-compatible endpoint. This is a standalone competition project rather than a general MCP server. 1. Extract the source and read the root and track_2b documentation. 2. For Docker, run make run. The first run separately downloads an approximately 5 GB model and needs the documented memory. 3. Alternatively configure an existing Apertus endpoint and use Python with your own date/value CSV. ```text make run ``` Code is Apache-2.0; documentation is CC-BY-4.0. The source ZIP does not include the separately downloaded model or runtime environment. Only data and research tools. Not investment advice. 只提供数据和研究工具，不构成投资建议。 Source and archive checks do not establish complete runtime or feature verification. --- # TradingAgents-astock - Name: TradingAgents-astock - Collection: Simon open-source skills and tools - Type: Workbench / framework - Author: Simon Lin - Source: https://github.com/simonlin1212/TradingAgents-astock - README: https://github.com/simonlin1212/TradingAgents-astock/blob/e3981af1e4c94656cdc663f1466d54fd49253dd5/README.md - Source commit: e3981af1e4c94656cdc663f1466d54fd49253dd5 - License: Apache-2.0 - Version: v0.5.21 - Formal release: yes - Release date: 2026-10-11 - Source verified: 2026-10-21 (Asia/Singapore) - Download: https://agentskill.nz/downloads/TradingAgents-astock-v0.5.21.zip - Bytes: 3876980 - SHA-256: f6db1e10e3a602c29f34bb37e3e80ab807c85e0a59b87c8edbbea6be86441e41 - Chinese page: https://agentskill.nz/projects/tradingagents-astock/ - English page: https://agentskill.nz/en/projects/tradingagents-astock/ ## 中文 将 TradingAgents 多 Agent 研究框架适配到 A 股的教学与研究实现。 - 组合市场、基本面、新闻与 A 股特化分析角色。 - 组织多空观点与风险讨论，用于研究流程实验。 - 提供本地 Web 界面与命令行入口。 支持与接入：框架使用用户配置的模型供应商；支持 Claude Agent SDK 可选接入，不是通用 Agent Skill。 1. 下载解压，准备 Python 3.10 或以上版本并创建虚拟环境。 2. 安装项目，按 README 配置自己的模型供应商。 3. 启动 Streamlit 界面；其他模式见 README。 ```text python -m pip install -e . streamlit run web/app.py ``` 基于 TauricResearch/TradingAgents 的改造版。原始 LICENSE 与归属信息保留，仅供研究与教学。 ## English An A-share adaptation of the TradingAgents multi-agent framework for research and teaching. - Combines market, fundamentals, news and A-share-specific analyst roles. - Structures bull/bear and risk discussions for research experiments. - Provides a local web interface and command-line entry point. Agents and compatibility: Uses model providers configured by the user, with optional Claude Agent SDK integration. It is a framework rather than a general-purpose agent skill. 1. Extract the source, prepare Python 3.10 or newer and create a virtual environment. 2. Install the project and configure your model provider as described in the README. 3. Start the Streamlit interface; the README documents other modes. ```text python -m pip install -e . streamlit run web/app.py ``` Adapted from TauricResearch/TradingAgents. Original license and attribution remain in the archive. For research and teaching only. Only data and research tools. Not investment advice. 只提供数据和研究工具，不构成投资建议。 Source and archive checks do not establish complete runtime or feature verification. --- # vibe-astock - Name: vibe-astock - Collection: Simon open-source skills and tools - Type: Workbench / framework - Author: Simon Lin - Source: https://github.com/simonlin1212/vibe-astock - README: https://github.com/simonlin1212/vibe-astock/blob/7988c991eaac0adee8b49bf3ee204f6a21e437aa/README.md - Source commit: 7988c991eaac0adee8b49bf3ee204f6a21e437aa - License: Apache-2.0 - Version: v2.1.2 - Formal release: yes - Release date: 2026-10-18 - Source verified: 2026-10-21 (Asia/Singapore) - Download: https://agentskill.nz/downloads/vibe-astock-v2.1.2.zip - Bytes: 11013683 - SHA-256: 481043d0282becbb38c2a09d714f97bd8cffb9591c013ccc34eab6d30839b694 - Chinese page: https://agentskill.nz/projects/vibe-astock/ - English page: https://agentskill.nz/en/projects/vibe-astock/ ## 中文 面向 A 股市场复盘、跟踪与研究记录的本地工作台。 - 查看盘面异动、市场情绪与题材轮动数据。 - 结合自己的 AI 整理复盘、周报与研究。 - 管理交易日志、跟踪记录和相关资料。 支持与接入：Claude Code、Codex、WorkBuddy / CodeBuddy 订阅，或兼容 Chat Completions 的模型 API。 1. 下载源码 ZIP，按 README 准备 Node 与 Python。 2. 在解压后的根目录运行 setup 与 start；Windows 使用对应 cmd 脚本。 3. 打开本地工作台，再接入自己的 AI。 ```text bash scripts/setup bash scripts/start ``` 本站只提供源码，未搬运安装包。数据口径与平台验证范围见对应版本说明。 ## English A local workbench for reviewing, tracking and documenting China A-share markets. - Displays market moves, sentiment and theme-rotation data. - Uses your own AI for daily reviews, weekly notes and research. - Manages trading journals, tracking records and supporting materials. Agents and compatibility: Claude Code, Codex, WorkBuddy / CodeBuddy subscriptions, or Chat Completions-compatible model APIs. 1. Download the source ZIP and prepare Node and Python following the README. 2. Run setup and start from the extracted root; use the matching cmd scripts on Windows. 3. Open the local workbench and connect your own AI. ```text bash scripts/setup bash scripts/start ``` Only source code is distributed here. Read the selected version's documentation for data conventions and platform verification limits. Only data and research tools. Not investment advice. 只提供数据和研究工具，不构成投资建议。 Source and archive checks do not establish complete runtime or feature verification. --- # Vibe-Research - Name: Vibe-Research - Collection: Simon open-source skills and tools - Type: Workbench / framework - Author: Simon Lin - Source: https://github.com/simonlin1212/Vibe-Research - README: https://github.com/simonlin1212/Vibe-Research/blob/94f61bf517393f80e88e974c224ac20b774ff990/README.md - Source commit: 94f61bf517393f80e88e974c224ac20b774ff990 - License: MIT - Version: v2.2.2 - Formal release: yes - Release date: 2026-10-18 - Source verified: 2026-10-21 (Asia/Singapore) - Download: https://agentskill.nz/downloads/Vibe-Research-v2.2.2.zip - Bytes: 8340837 - SHA-256: 84848199ccf11bba3c3fc72b9846b14569a9773fd05cc96dc3b5c3e4bd1ac016 - Chinese page: https://agentskill.nz/projects/vibe-research/ - English page: https://agentskill.nz/en/projects/vibe-research/ ## 中文 将市场复盘、产业研究、公司研究与资料库放在一起的本地工作台。 - 整理 A 股、美股和港股的市场与公司数据。 - 分阶段开展公司研究，保留研究记录、来源与计算。 - 提供资料检索、历史回测、自选与持仓记录。 支持与接入：Claude Code、Codex、WorkBuddy / CodeBuddy 订阅，或兼容 Chat Completions 的模型 API。 1. 下载并解压源码，按 README 准备 Node 与 Python。 2. macOS / Linux 运行 setup 后运行 start；Windows 使用对应 cmd 脚本。 3. 在本地页面的「接入 AI」里配置自己的订阅或 API。 ```text bash scripts/setup bash scripts/start ``` 本站提供源码，不提供大型安装包。系统兼容性与具体模型的验证范围以对应版本 README 为准。 ## English A local workbench for market reviews, industry and company research, and source materials. - Organizes market and company data for China A-shares, US and Hong Kong markets. - Runs staged company research while preserving records, sources and calculations. - Includes document retrieval, historical backtesting, watchlists and portfolio records. Agents and compatibility: Claude Code, Codex, WorkBuddy / CodeBuddy subscriptions, or Chat Completions-compatible model APIs. 1. Extract the source and prepare Node and Python as specified in the README. 2. On macOS / Linux run setup, then start; Windows uses the matching cmd scripts. 3. Configure your own subscription or API in the local Connect AI screen. ```text bash scripts/setup bash scripts/start ``` This site distributes source code rather than large installers. Check the selected README for platform and model verification limits. Only data and research tools. Not investment advice. 只提供数据和研究工具，不构成投资建议。 Source and archive checks do not establish complete runtime or feature verification.