Skip to main content

superpowers-dsh

72Stars2Forks0Issues0Watchers

Ports 14 AI collaboration skills from obra/superpowers to DeepSeek Harness: TDD, debugging, planning, sub-agent dispatching, and more. Ready to use out of the box.

Evidence4/5methodologySourceInstallMaintenanceDSH versionSecurity scan
Machine-auditedInstall commandRepo verifieddsh-plugin topicLicenseREADMEAI wiki
Language
JavaScript
Branch
main
deepseek-harnessdsh-pluginskillssuperpowers

Install

cmdweb profile
$ dsh plugin --profile web add superpowers-dsh

Run the command above in your terminal to install this plugin via the dsh CLI. You can switch Profile in the top-right corner. New to dsh? Read the beginner tutorial

Install via your agent

Install the DeepSeek Harness plugin LayneChai/superpowers-dsh for me: review the repository at https://github.com/LayneChai/superpowers-dsh first, then run the install command and verify the plugin loads successfully.

Paste this instruction to the DSH Web GUI assistant — it will install and verify for you.

One-Line Description

A collection of 14 AI collaboration skills packaged for DeepSeek Harness (DSH): transplanting the obra/superpowers skill library (TDD, debugging, planning, sub-agent dispatch, etc. from Claude-Code) to DSH's Cordis plugin architecture as-is, so all agent sessions can see and load them from the skills directory after installation.

Core Capabilities

  • Register 14 DSH skills, covering brainstorming, planning execution, sub-agent dispatch, TDD, systematic debugging, code review, worktree isolation, and other workflows
  • Map upstream Claude-Code tools to DSH equivalents (subagent, subagent_fork, workflow, goal, etc.) without modifying call patterns
  • Attach skills to the host registry through the Cordis bundle layer, with each agent preset's scope chain automatically merged
  • Auto-discover skills/<name>/SKILL.md: dropping new files into the package adds new skills without code changes
  • list() scans directories only, get() reads content on demand—lazy loading avoids parsing all skills at once
  • Return a directory-type resourceBase for each skill, enabling correct path resolution for scripts and prompt templates referenced within skills

Technical Implementation

  • Language: JavaScript (ESM)
  • Key Dependencies: No runtime dependencies (only three built-in modules: node:fs/promises, node:url, node:path)
  • Architecture Pattern: Cordis plugin + ctx.skills.registerProvider(), inserting a line via cordis.patch.yml on top of the dsh-base layer (id: superpowers-dsh) as a host-plane bundle; list() returns candidates, get() loads SKILL.md content on demand
  • Entry File: lib/index.js

Use Cases

When you want DSH agents to brainstorm before coding, skip root cause analysis during debugging, run code review before commits, split complex tasks to sub-agents for parallel processing—but don't want to write prompts manually for each project, installing this plugin is enough. It's suitable for medium-to-large projects requiring long-term maintenance, especially where teams have high engineering discipline standards but inconsistent prompts.

Prerequisites & Compatibility

DependencyMinimum VersionNotes
DeepSeek Harness (DSH)Not declaredMust register via dsh plugin --profile web add command; no specific version declared in package.json
Node.jsNot declaredOnly uses built-in modules; no minimum Node version declared
PlatformCross-platformPlugin uses Node built-in modules; runs on macOS/Windows/Linux
Native ModulesNoneNo node-gyp / native dependencies

Installation

dsh plugin --profile web add github:LayneChai/superpowers-dsh

Configuration

No additional configuration required. Skill content, skill list, and registration behavior are entirely determined by files within the package—no environment variables, config files, or Schema fields needed.

FAQ

Q: Why can't I see the skills after installation?

A: The plugin layer mounts at profile startup. After installation, you must restart dsh web (or npx @deepseek-ai/dsh web) and refresh the browser, then use dsh --profile web --dump-config to check if there's a superpowers-dsh line in the output.

Q: Does it conflict with DeepSeek Harness's built-in skills?

A: No conflict. It registers with source: 'custom', rank: 550, lower than the built-in root skill directory; scope chains are merged rather than overridden.

Q: Can I add or modify skills?

A: Yes. When installed from a folder or file: spec, it's a link installation. Drop a new file into skills/<kebab-name>/SKILL.md (with YAML frontmatter: name + description, optional whenToUse), and it will be auto-discovered on next profile restart—no JS code changes needed.

Q: Does this plugin read/write local data or make network calls?

A: No. It only reads SKILL.md files from its own package and injects skill definitions into the host skill registry—no persistent storage, no extra network requests.

Q: Does it support Windows?

A: The plugin body uses Node built-in modules and runs fine on Windows; however, the .sh helper scripts bundled with the brainstorming skill are bash-specific and won't work on Windows native command line—WSL or Git Bash is needed.

Q: Do I need an npm account or 2FA to install?

A: No, it's just a regular package download. After publishing to npm, it will sync to npmmirror in China.

Q: What's the difference from upstream obra/superpowers?

A: Three main differences: removed namespace prefix (superpowers:brainstorming → brainstorming); rewrote using-superpowers to hook into DSH's skill tool; replaced Claude-Code specific tools with DSH's subagent / subagent_fork, etc.

Q: How do I uninstall?

A: Run dsh plugin --profile web remove superpowers-dsh, then restart the profile.

Difficulty

Beginner — one install command, one restart, and it works without modifying any project files; to customize skills, just add markdown under skills/, zero code barrier.

Known Issues & Limitations

  • The .sh visual companion script bundled with the brainstorming skill is bash-only and cannot run on Windows native shell (use Node version scripts/server.cjs or WSL/Git Bash on Windows)
  • Plugin rank is fixed at 550 and cannot be adjusted via user config—if there's a conflict with a future same-named skill with higher rank, you need to modify lib/index.js and repackage
  • YAML frontmatter parsing only covers scalar fields (name / description / whenToUse); lists or nested structures won't be expanded, but other fields are passed through as metadata

Read the usage guide →

Install steps, key points, FAQ and compatibility for this plugin — auto-derived from indexed fields.

Listing badge

Listed on deepseek-plugin.org
[![Listed on deepseek-plugin.org](https://img.shields.io/badge/listed_on-deepseek--plugin.org-007EC6)](https://deepseek-plugin.org/plugins/LayneChai/superpowers-dsh)

Paste this markdown into your GitHub README to link back to this listing. The badge only states the listing — not a security endorsement.

← Back to plugin directory