AI Agent Hub
Back to plugins
🖥️

dsh-skill-station

Client Updated 2026.08.25

Run the following command in DeepSeek Harness:

dsh plugin install WilShi/dsh-skill-station

Paste the following prompt into your AI chat to install this plugin:

Run dsh plugin install WilShi/dsh-skill-station in your terminal; the source lives at https://github.com/WilShi/dsh-skill-station , then restart dsh web to see the Skill Station button in the sidebar.

About this plugin

Managing AI skills inside DeepSeek Harness has often meant typing terminal commands, copying files between folders by hand, and eyeballing frontmatter for formatting mistakes. dsh-skill-station collapses all of that into a single sidebar button: open the panel and you can browse global, shared, and per-workspace skill roots with search, scene tags, enable/disable toggles, and restorable deletion.

Its real strength is cross-tool interoperability. The plugin performs a read-only scan of skill directories from Claude Code, Codex CLI, Cursor, and Gemini CLI, then lets you pick candidates and import them into Harness writable roots with a conflict policy of skip, rename, or replace. You can also drag a skill folder or zip archive straight onto the install tab, or paste an absolute local path for a disk-level copy that easily carries large skills with vendored dependency trees. For existing skills, the panel offers a file-tree browser, in-place SKILL.md editing, exportable re-importable zips, and a built-in diagnostics pass that flags strict-YAML violations, missing fields, and legacy keys with one-click repair.

It is built for developers who juggle multiple AI CLI tools and want to centralize scattered prompt skills into Harness for unified management. The security model confines all writes to the listed writable roots, containment-checks every path after normalization, never follows symlinks during copies, keeps external agent directories strictly read-only, and rejects path-traversal segments in uploads. MIT licensed, every config field is optional, and it works out of the box.

Use Cases

  • Import scattered prompt skills from Claude Code and Cursor into Harness for unified management
  • Drag a zip archive onto the install tab to deploy a large skill folder with vendored dependencies
  • Batch-repair frontmatter formatting errors via the diagnostics pass, then re-enable affected skills

Best For

  • Developers juggling multiple AI CLI tools who want centralized prompt skill management
  • Engineering leads maintaining shared skill libraries who need batch import, export, and repair
  • Tech users who want to safely onboard third-party skills into Harness without modifying source directories