dsh-task-checklist
by fff122
DeepSeek Harness 的本地任务清单插件
Local task checklist plugin for DeepSeek Harness.
安装
dsh plugin --profile web add github:fff122/dsh-task-checklistGitHub 源码安装:首次需按提示配置 allowBuilds 构建授权后重试
安装与环境配置指引、插件开发教程见 DSH 中文社区文档 ↗
安装即在你的机器上以你的权限运行第三方代码——它可读写文件、使用凭据、访问网络,DSH 的工具审批不会为插件代码加沙箱。「检测到 manifest」仅代表发现 dsh.bundle / dsh.plugin 清单,不构成兼容性或安全审查;安装前请审阅源码,不熟悉的插件先在不含密钥的环境试用。
README
目录
A local task checklist plugin for DeepSeek Harness. It lets an Agent create tasks, filter them by state or tag, mark work complete, and export a clean Markdown checklist.
All task data stays in the current working directory under .dsh/task-checklist/tasks.json. The plugin does not send task data over the network.
Quick install
Requirements
You need Node.js 22 or newer, pnpm, and a DeepSeek Harness installation that supports JavaScript or TypeScript plugins.
Step 1: Download and build
Copy and run:
git clone https://github.com/fff122/dsh-task-checklist.git
cd dsh-task-checklist
pnpm install
pnpm build
Step 2: Create the plugin configuration
Run this command from the repository root:
cat > task-checklist.patch.yml <<EOF
- insert:
- id: dsh-task-checklist
name: '$PWD/dist/src/index.js'
EOF
The plugin path is absolute because the Harness patch loader requires an absolute path.
Step 3: Start Harness with the plugin
If Harness is installed as a command:
dsh web --patch "$PWD/task-checklist.patch.yml"
Otherwise, run:
npx @deepseek-ai/dsh web --patch "$PWD/task-checklist.patch.yml"
Verify the installation
Ask the Agent:
创建一个任务:准备周五的项目演示。标签用 work 和 demo,优先级为 high。
If the plugin is loaded, the Agent can use task_create. You can then ask it to list unfinished tasks, complete a task by id, or export a Markdown checklist.
Available tools
| Tool | Purpose |
|---|---|
task_create |
Create a task with a title and optional details, tags, and priority. |
task_list |
List tasks, optionally filtered by todo or done status and matching tags. |
task_complete |
Mark a task complete using its id. |
task_export_markdown |
Export matching tasks as a Markdown checklist. |
Data and behavior
Each Harness working directory has its own checklist at .dsh/task-checklist/tasks.json. Writes are atomic, so an interrupted write cannot leave a partially written data file. Task identifiers are returned by task_create and task_list; use them with task_complete.
The tag filter is an all tags filter. For example, filtering by work and demo returns only tasks that have both tags.
Updating the plugin
From the cloned repository:
git pull
pnpm install
pnpm build
Restart Harness with the same patch command after rebuilding.
Development
Run the complete package quality gate from the repository root:
pnpm run format:check
pnpm run typecheck
pnpm run test
pnpm run build
The implementation is deliberately separated into small modules:
src/
index.ts # Harness entry point and tool registration
task-schema.ts # Task types and input normalization
task-store.ts # Local persistence, filtering, completion, and export
test/
index.test.ts
mount.test.ts
task-schema.test.ts
task-store.test.ts
For the plugin model and patch overlay format, see the official DeepSeek Harness plugin guide.
原始 README: https://github.com/fff122/dsh-task-checklist/blob/main/README.md ↗
同类插件
查看全部 →
deepseek-harness
从仓库或系统描述生成经过校验的自包含交互式架构图、流程图、时序图、数据流图和生命周期图。

dsh-plugin
通过 DSH MCP 客户端挂载 Ouroboros 的纯配置包,在 DSH 中提供 36 个涵盖需求访谈、Seed、执行、评估与演化流程的工具。

dsh-tongflow
基于 TongFlow 的“片场”插件,用于图片、配音、音乐与视频制作:agent 为每个资产生成 TongFlow 工作流文件(.tongflow.json)并通过 TongFlow 插件执行,内嵌工作流画布,按镜头/角色/take 组织项目,附漫剧模板;以 @tongflow 开头的会话进入 Studio 界面。

helloagents
AI 编码 CLI 的工作流层:技能、项目知识、交付检查、更安全的配置写入与可恢复执行

dsh-ai-novel-writer
安装专用 AI 小说创作预设与工作台:提供带修订号的本地项目资产、紧凑侧边工作台,以及需要原生审批的逐文件变更。

rea
用 agent 逆向任何东西:从应用行为到原生二进制