Skip to content
View zinan92's full-sized avatar
🎯
Focusing
🎯
Focusing
  • shanghai
  • 04:37 (UTC -12:00)

Highlights

  • Pro

Block or report zinan92

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.

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
zinan92/README.md

Park

I build AI-agent operating systems for markets, content, and product creation.

用 AI Agent 把交易、内容与产品创造,变成可以持续运行和学习的系统。

上海 · Builder · Systems thinker

Park's three operating systems: Trading OS, Content OS, and Agent Product Lab

The full picture

我不是在收集互不相干的 repo。我在搭三套相互连接的操作系统:

System Outcome Current focus
Trading OS 从市场事实走到受约束的交易决策与复盘 Paper-first 闭环与可审计证据
Content OS 从内容信号走到跨平台成品与反馈学习 独立 capability 串成生产系统
Agent Product Lab 从产品意图走到真实使用证据 Build → ready gate → Use 的交接闭环

READY 可用入口与核心结果明确 · BUILDING 正在形成完整产品 · EXPLORING 仍在验证方向


Trading OS

Outcome: 把多市场数据转成有来源、有风控、有执行证据的决策闭环。当前坚持 paper-first;不在这里宣称真钱能力已经开放。

System map

Market facts          Understanding          Decision             Learning

01 Data        ──▶ 02 Intelligence ──▶ 03 Signal       ──▶ 04 Method routing
   READY              READY                BUILDING          BUILDING
      │
      ▼
05 Backtest    ──▶ 06 Risk plan    ──▶ 07 Paper execution ──▶ 08 Journal
   READY              BUILDING             BUILDING              EXPLORING

Products

Product Role in the system Status
datafeed ticker + timeframe → multi-market OHLCV READY
intel 10+ sources → scored, clustered market events READY
equity-research evidence snapshots → A-share investment-committee research BUILDING
backtest strategy definition → win rate, payoff and drawdown READY
standard-kline OHLCV + provenance → trustworthy chart surface READY

Now / Next

  • Now — 强化从行情、情报、策略到 Paper 执行的可审计闭环;把“代码存在”和“真实运行证据”分开。
  • Next — 用完整周期证据验证多资产闭环,再决定哪些能力值得进入更高风险阶段。
More context
  • quant-data-pipeline 是早期多市场数据与信号能力的集成底座。
  • 私有执行与风控实现不会从 Profile 暴露;公开页面只描述能力边界和已验证状态。
  • Backtest 是辅助证据,不自动等于策略可交易,更不等于 live-ready。

Content OS

Outcome: 让创作者负责判断与表达,让系统处理发现、获取、理解、生产、组装、分发和反馈。

System map

Discover             Understand             Create                Learn

01 Signals     ──▶ 02 Acquire       ──▶ 03 Extract      ──▶ 04 Curate
   BUILDING           READY                 READY                EXPLORING
      │
      ▼
05 Rewrite     ──▶ 06 Assemble      ──▶ 07 Publish      ──▶ 08 Performance
   READY              READY                 BUILDING             EXPLORING

Products

Product Role in the system Status
content-intelligence social data → trends, patterns and topic signals BUILDING
content-downloader platform URL → normalized media + metadata READY
content-extractor video / image / article → structured text READY
content-rewriter source material → platform-specific drafts READY
videocut talking-head footage → edited video assets READY
daily-newsletter source feeds → selected Chinese daily brief + receipts READY

Now / Next

  • Now — 独立能力已经覆盖获取、理解、改写和视频组装;重点是让它们以清晰合同协作,而不是继续堆工具。
  • Next — 补齐 curator 与 performance feedback,让选题质量和发布结果能够回流到下一轮生产。
More context
  • seedance-expert 把视频创意转成可执行的多模态生成提示。
  • AI-videos 探索虚拟人物换装与动作迁移工作流。
  • 已归档的 orchestrator 和 workbench 保留为历史证据,不再占据主地图。

Agent Product Lab

Outcome: 把独立产品想法做成 ready for use 的产品,再用真实任务验证价值,把缺口送回 Build。

System map

Intent              Build                 Gate                 Use

01 Explore    ──▶ 02 Product build ──▶ 03 Readiness     ──▶ 04 Real tasks
   READY             BUILDING             READY                BUILDING
                                                                  │
                                                                  ▼
                   06 Improve      ◀── 05 Evidence
                      READY              READY

Products

Product Role in the system Status
proactive-explorer existing product → evidence-backed next direction READY
doc-driven-dev-workflow intent → reviewable development stages and guards READY
wechat-miniprogram-shipping product intent → release contract and evidence path READY
repo-evals product claims → reproducible verdict dossier READY
loop repo + contract → value-ranked issues, PRs and digest READY
codex-harness local agent sessions → project and token evidence BUILDING

Now / Next

  • Now — Product Lab 已明确分成 BuildUse:Build 对 readiness 负责,Use 对真实任务与使用证据负责。
  • Next — 把 ready handoff、真实使用、缺口复现和回流需求做成跨产品可复用的证据链。
Operating rule
Intent → Issue contract → Build → Readiness gate → Real use → Evidence → Next issue

绿测试证明代码通过了测试,不自动证明产品 ready;部署成功也不自动证明用户结果已经发生。


How the systems connect

                 ┌─────────────────────────┐
                 │    Agent Product Lab    │
                 │  builds + tests + learns│
                 └────────────┬────────────┘
                              │
                     product capabilities
                              │
             ┌────────────────┴────────────────┐
             ▼                                 ▼
       ┌────────────┐                    ┌────────────┐
       │ Trading OS │                    │ Content OS │
       │ decisions  │                    │ production │
       └─────┬──────┘                    └─────┬──────┘
             └──────────── evidence ───────────┘
                              │
                              ▼
                       better next build

三套系统共享同一条原则:先定义结果,再建立可复验合同;系统必须诚实表达 READY、BUILDING 和 EXPLORING。

Working together

如果你也在构建 agent-native 产品、研究系统或内容基础设施,可以从相关 repo 的 issue 开始交流。最好的合作入口不是“聊一个大想法”,而是一个清楚的问题、输入、期望输出和失败边界。

Build the system. Use the system. Keep the evidence.

Pinned Loading

  1. agent-core agent-core Public

    AI agent 操作系统内核 — architecture-first starter package,提供 onboarding、skills、SOPs、runtime specs、curated knowledge 五大原语

    Shell 1

  2. intel intel Public

    情报采集。in 10+信息源 → out LLM评分+跨源事件聚类

    Python 1

  3. quant-data-pipeline quant-data-pipeline Public

    多市场量化数据平台 — A股/美股/加密/商品,28组API,感知信号引擎,模拟交易

    Python 2

  4. videocut videocut Public

    AI 口播视频编辑。in 视频文件/目录 → out 去废话+字幕+金句+拆条+封面+变速

    JavaScript 15 2