dsh-skill-creator
Run the following command in DeepSeek Harness:
dsh plugin install TheHeartFickle/dsh-skill-creator
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install TheHeartFickle/dsh-skill-creator in your terminal to install the plugin; the source code is hosted at https://github.com/TheHeartFickle/dsh-skill-creator.
About this plugin
Building a production-quality DSH skill typically means interviewing, drafting, evaluating, reviewing, and iterating, yet the supporting tooling is often scattered, hard to quantify, and feedback collection remains largely manual. dsh-skill-creator packages the entire pipeline into a ready-to-use DSH plugin that walks you through requirements interviews, draft generation, automated benchmarking, human review, and iterative refinement in a single, reproducible local loop.
Under the hood the plugin ships a pure Node.js CLI (validate, benchmark, review) with zero third-party dependencies. After a run completes, a bundled local web viewer renders quantified scores, per-item grades, and structured feedback, so you no longer need to scroll through raw JSON. All generated artifacts, including skill drafts, evaluation workspaces, benchmark results, and review feedback, are written only to your chosen workspace or the system temp directory, leaving install locations such as .dsh/skills completely untouched.
Whether you are scaffolding your first skill, running regression benchmarks on an existing one, or enforcing a team-wide quality gate, dsh-skill-creator provides a lightweight, auditable workflow foundation. The current release includes one fully functional skill validated by 27 passing test cases, with plans to extend skill-development capabilities while keeping the toolchain dependency-free and easy to maintain.
Use Cases
- Go from requirement interviews to skill drafting, benchmarking, and iteration in one flow
- Run regression benchmarks on an existing skill and quantify the delta
- Review quantified scores and structured feedback in a local web viewer instead of raw JSON
Best For
- Developers and teams building or maintaining DSH skills
- Independent developers who prefer a zero-dependency, auditable toolchain
- Skill architects enforcing quantified evaluation and a unified quality gate
Related Plugins
A method pack that makes AI coding agents plan against your real baseline, prove completion with fresh evidence, and reduce reworks and unsafe changes.
Turns the DeepSeek Harness session into a captain that builds a durable sub-agent team, splits goals into dependency-aware tasks, and coordinates work via direct messages and a live Web UI.
Gives coding agents design judgment, letting Claude Code, Cursor, and 70+ agents generate and iterate high-quality UI, presentations, and graphics right from the terminal.
Run the Pi ecosystem's plugins on DeepSeek Harness, unmodified, via a compatibility layer that implements Pi's public extension ABI on DSH's native services.