Preface¶

In the DSH ecosystem, there are numerous sub-tool plugins with scattered naming, making them hard to recognize at a glance in the community directory. When creating automation scripts or agent tasks, there is often a need for basic capabilities like date conversion, encoding/decoding, JSON queries, CSV parsing, regex processing, Markdown conversion, diff comparison, statistical calculations, and schema validation. If each tool requires separate plugin discovery, installation, and configuration, the setup cost becomes high, and there is a risk of missing dependencies.

omdsh-dev/dsh-toolkit consolidates these ten tools into a single collection repository in vendored snapshot form, providing unified engineering, testing, and batch installation scripts. Each sub-package can still be installed, enabled, disabled, or uninstalled independently. For scenarios requiring all tools at once, a meta-package model is also available. This section introduces its positioning, capabilities, and installation methods.

What It Is¶

dsh-toolkit is maintained by omdsh-dev and classified as a client-side plugin on SkillHub. The current version is v0.0.1 under the MIT license. Repository: github.com/omdsh-dev/dsh-toolkit; Community directory page: skillhub.cn/plugins/omdsh-dev/dsh-toolkit.

Its official positioning is a collection and installation assistance repository: the sub-package source repositories evolve independently (under the omdsh-dev organization), while this repository freezes pack artifact snapshots and retains @deepseek-ai/dsh-toolkit as an optional atomic-mount meta-package. The root meta-package is marked private: true for Git/collection distribution and is not intended for publication to the npm registry.

Ten Tools and Their Capabilities¶

The repository README lists the following sub-tools along with their test case counts (totaling 723 cases):

Each tool emphasizes zero dependencies and deterministic output, making them suitable for reproducible data processing in agent pipelines without relying on external networks or dynamic evaluation.

Two Runtime Models¶

The README distinguishes two mounting methods; choose one based on the scenario.

Standalone Bundle Model (Recommended): Each sub-package is registered individually, allowing installation of only the needed tools.

# Install a single tool to the web profile
dsh plugin --profile web add github:omdsh-dev/dsh-tool-csv

# Install to the headless profile (default for dsh run)
dsh plugin --profile headless add github:omdsh-dev/dsh-tool-diff

Meta Bundle Model (Optional): Atomically mount all ten tools at once.

dsh plugin --profile web add github:omdsh-dev/dsh-toolkit
dsh --profile web --dump-config | grep tool-kit

Note: If the profile already has individually mounted sub-plugins of the same name (e.g., tool-time, tool-csv), mounting the meta-package will cause a naming conflict error. Remove the old plugins first or use the standalone bundle model throughout. The meta-package’s apply is atomic—any sub-plugin registration failure triggers a reverse rollback without leaving partial state.

Installation and Enablement¶

The web and headless are different profiles: installing in web does not automatically override headless; dsh run defaults to headless. On Windows paths use forward slashes (C:/...).

Batch Installation (Collection Script)¶

After cloning or entering the repository, use auxiliary scripts for idempotent batch installation (repeated execution does not duplicate additions):

./scripts/install-web.sh       # All 10 tools → web profile
./scripts/install-headless.sh  # All 10 tools → headless profile
./scripts/install-all.sh       # Install to both profiles

Verify Installation¶

dsh --profile web --dump-config | grep tool-csv     # Installation successful if line exists
dsh run "使用 csv 工具解析 'a,b\n1,2'"               # headless end-to-end test

Local Tarball Installation¶

Without relying on GitHub, you can build locally and install via npm pack artifacts:

npm pack
dsh plugin --profile web add <path to npm pack artifact tarball>

For the headless profile, replace web with headless.

Runtime Environment¶

package.json requires Node ^22.19.0 || >=24.0.0 with packageManager set to npm@11.16.0. In standalone build mode, after running npm install, you can execute npm run build:all and bash scripts/test-all.sh for local validation.

Typical Usage¶

After installing the CSV tool standalone, you can call it end-to-end via headless:

dsh run "使用 csv 工具解析 'a,b\n1,2'"

Install only the diff tool as needed for text comparison:

dsh plugin --profile headless add github:omdsh-dev/dsh-tool-diff

When requiring the full toolset and with no conflicting sub-packages in the profile, mount the meta-package:

dsh plugin --profile web add github:omdsh-dev/dsh-toolkit

Use Cases and Notes¶

Who It’s For

• Those needing basic operations like dates, encoding, JSON, CSV, regex, Markdown, diff, statistics, and schema in DSH agents or scripts, with deterministic and testable outputs.
• Those wanting to view the list and version snapshots of ten tools in one collection and batch install them into web or headless profiles via scripts.
• Those preferring the standalone bundle model for installing only one or two sub-tools while keeping others minimal.

Before Use

• Plugins run with the current dsh process permissions; before installation, review the source code and MIT license to ensure compliance with your security policies.
• SkillHub is a community directory site, not officially affiliated with DeepSeek or High-Flyer, and does not equate to an official app store.
• Sub-packages evolve in their respective source repositories; this collection is a vendored snapshot. See docs/CONTRIBUTING.md and the vendored sync notes in the README for update synchronization.

Conclusion¶

dsh-toolkit consolidates ten commonly used zero-dependency tools from the DSH ecosystem into a single collection, reducing configuration costs through unified testing (723 test cases) and installation scripts while retaining the flexibility to manage sub-packages independently. If you are building a DSH toolchain, you can explore the overview on the directory page, check the full README and catalog.json on GitHub, and install as needed.

• Directory page: https://www.skillhub.cn/plugins/omdsh-dev/dsh-toolkit
• GitHub: https://github.com/omdsh-dev/dsh-toolkit