dsh-plugin-factory
Run the following command in DeepSeek Harness:
dsh plugin install ktziy/dsh-plugin-factory
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install ktziy/dsh-plugin-factory in your terminal to install; the full source is available at https://github.com/ktziy/dsh-plugin-factory
About this plugin
DeepSeek Harness enforces a strict plugin contract: the apply(ctx) lifecycle, defineTool definitions, and hard Schema DSL constraints. Most LLMs were never trained on this API surface, so asking an agent to write a plugin from scratch often produces hallucinated fields and invalid calls that fail at runtime.
dsh-plugin-factory packages the contract into three practical tools. plugin_doc retrieves documentation and official references by topic or file path. plugin_scaffold generates a ready-to-compile skeleton for six plugin kinds: tool, service, event, llm-adapter, client, and client-node. plugin_validate runs guard-rule static checks before the code ever reaches the runtime, catching violations early. Together they form a docs-to-scaffold-to-verify loop that keeps every agent output grounded in the real spec.
If you prefer driving engineering tasks with natural language and do not want to re-read the official docs and hand-check every API for each new plugin, this is a lightweight way to inject spec knowledge directly into the conversation. Describe what you need in plain language and receive back a compilable skeleton plus a paste-ready install snippet.
Use Cases
- Have the agent in a dsh session auto-generate valid plugin skeleton code
- Review contract docs and validate API compliance before shipping a new plugin
- Quickly generate compilable templates for tools, services, and LLM adapters
Best For
- Engineers who use dsh daily and want to drive plugin creation with natural language
- Developers new to dsh plugins who want to reduce API hallucinations
- Workflow users who need to batch-generate skeleton code across multiple plugin kinds
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.