Skip to content

TaskCarry 0.2.0rc4

Pre-release
Pre-release

Choose a tag to compare

View all tags
@bryan123466 bryan123466 released this 27 Oct 11:16
v0.2.0rc4
f130e31

Project memory and collaboration for AI tools.

TaskCarry keeps task progress, saved checkpoints, delivered stages and continuation notes with your project. Work in independent AI workspaces, bring ready results together, and continue supported interrupted operations using recorded and actual effects.

Download and start

Download taskcarry-0.2.0rc4-product.zip below, extract the product folder, and open START_HERE.md. The ZIP includes the installable wheel, four Skills, three-language manuals and examples. Python 3.11+ is required; Git is needed for workspace integration.

Start in a governance chat, connect the workspaces you choose, and request the first governance run. See its real result in that chat; optional scheduled governance uses the same tools and workflow. The concrete walkthrough uses Codex. CLI and MCP provide integration points for other local AI tools, whose native connection and behavior require separate verification.

This candidate consistently uses taskcarry for the package, command and MCP tools, and .taskcarry for project state. It includes refreshed English, Simplified Chinese and Traditional Chinese product documentation, TaskCarry directory examples, collaboration diagrams and a new sharing image.

Validation

  • Local source suite: 160 cases; 157 passed, 3 skipped, no failures or errors. Two skips require unavailable native symbolic-link privileges; the installed-wheel case passed separately.
  • The final downloaded-product shape was installed outside the source checkout. 47 walkthrough checks passed, including three workspaces, cumulative delivery, paused checkpoints, independent stages, recovery after a real Git effect, and repeated execution without a duplicate commit.
  • Four packaged Skills and real stdio MCP startup passed. All 82 packaged source/document members match the frozen Git content.
  • Version-specific validation · Public CI for this commit

Native client activation, real scheduled triggers and physical multi-computer operation were not exercised in this release validation. This is a release candidate. No package-registry publication is implied; install the wheel included here.

简体中文

TaskCarry 为本地 AI 工具提供项目记忆与工作区协作。下载下方 taskcarry-0.2.0rc4-product.zip，解压后打开 START_HERE.md，按手册建立治理聊天并接入自己的任务。首次治理直接在该聊天展示结果，后续定时治理复用同一流程，无需另装报告应用。

本版统一产品、命令、MCP、Skills 和示例目录名称；提供英文、简体、繁体操作说明，以及新的协作图与分享图。当前上手示例使用 Codex，其他本地 AI 工具可通过 CLI/MCP 适配，并需验证各自的实际接入。

源码 160 项验证中 157 通过、3 跳过，安装包验证另行通过；最终产品包的三工作区演练通过 47 个检查点。原生客户端激活、定时自然触发和实体多电脑不在本次已验证范围。使用下方安装包，无需用户运行开发测试。

繁體中文

TaskCarry 為本機 AI 工具提供專案記憶與工作區協作。下載下方 taskcarry-0.2.0rc4-product.zip，解壓縮後開啟 START_HERE.md，依手冊建立治理對話並接入自己的任務。首次治理直接在該對話顯示結果，後續排程治理重用相同流程，無須另裝報告應用程式。

本版統一產品、命令、MCP、Skills 與目錄範例名稱，提供英文、簡體、繁體操作說明及新的協作圖與分享圖。目前上手範例使用 Codex，其他本機 AI 工具可透過 CLI/MCP 適配，並須驗證各自的實際接入。

原始碼 160 項驗證中 157 通過、3 跳過，套件安裝驗證另行通過；最終產品套件的三工作區演練通過 47 個檢查點。原生用戶端啟用、排程自然觸發及實體多電腦不在本次已驗證範圍。使用下方安裝套件，無須使用者執行開發測試。

MIT licensed. SHA-256 checksums for both downloads are in SHA256SUMS.txt.

Assets 5
Loading