Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/w2112515/dsh-plugin-development/dsh-plugin-developmentnpx skills add w2112515/dsh-plugin-development --skill dsh-plugin-developmentgit clone --depth 1 https://github.com/w2112515/dsh-plugin-developmentWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/w2112515/dsh-plugin-development/dsh-plugin-development)<a href="https://agentmods.dev/skills/w2112515/dsh-plugin-development/dsh-plugin-development"><img src="https://agentmods.dev/badge/skills/w2112515/dsh-plugin-development/dsh-plugin-development.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00130 | $0.01118 |
| Opus 5 | $0.00065 | $0.00559 |
| Sonnet 5 | $0.00026 | $0.00224 |
| Haiku 4.5 | $0.00013 | $0.00112 |
Grade A, and why
dsh-plugin-development scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 51 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Develop and Audit DeepSeek Harness Plugins
Classify the plugin before designing it. Dynamic runtime packages, repository workspace packages, and installable bundles use different source formats, loaders, persistence models, and evidence. Do not transfer a rule from one mode without confirming that it applies.
Establish authority and intent
- Identify the requested outcome, target files or runtime, and whether the user asked to inspect or to change it.
- Locate the current DeepSeek Harness checkout and read authority and verification. Use current repository evidence instead of memorized package lists or copied tutorial manifests.
- Determine the plugin mode from the target and available tools. If the request is a marketplace solution pack (整合包,
dsh.pack.json,dsh-plugin-pack, or a curated catalog list), stop. Usedsh-marketplace-publish. Do not select a plugin mode or authordsh.bundle/cordis.patch.ymlfor that artifact.
| Mode | Strong indicators | Read next |
|---|---|---|
| Dynamic runtime | cordis_define, cordis_run, a live Host or Client, Slots, @pluginId, or a process-local customization |
runtime dynamic plugins |
| Workspace package | A target under packages/, a repository capability or Consumer, or a change shipped as part of DSH |
workspace plugins |
| Installable bundle | dsh.bundle, cordis.patch.yml, dsh plugin, profile composition, npm, a tarball, or an out-of-tree package |
installable bundles |
When a request intentionally crosses modes, split it into explicit phases and load each applicable reference. When the evidence remains ambiguous and the modes would produce different artifacts, ask one focused question: whether the result should be process-local, shipped in the DSH repository, installable into a profile, or a marketplace solution pack (then switch to dsh-marketplace-publish).
What ships with it
7 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 3d ago First seen · 51 lines · 130 tokens per session scan A 9511ff5ee6c4
dsh-plugin-development is a skill published in the GitHub repository w2112515/dsh-plugin-development (14 stars, last pushed 17d ago), licensed MIT. It adds 130 tokens to every session and 1,118 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
dsh-plugin-guide
Use when developing, reviewing, packaging, debugging, or answering questions about DeepSeek Harness (DSH) plugins — the plugin-based agent harness on vendored Cordis. Applies the official plugin-development constraints (plugin contract, cordis.yml layers, services/events/effects, tool DSL, bundles/profiles) backed by…
codex-sync
Operate dsh-codex-sync from this session — preview and import Codex chats, toggle Skills/MCP/import settings, check MCP mirror health, install the reverse MCP bridge.
development-cdp
Inspect and automate the DeepSeek Harness Web UI through Chrome DevTools Protocol (CDP). Use when verifying dsh-pihuo plugins in the browser, checking startup, composer, or tool cards, taking screenshots, or running scripts/cdp-check.mjs. Triggers: CDP, Chrome remote debugging, pageinfo, "网页调试", /development-cdp.
dsh-plugin-author
Scaffold and implement DeepSeek Harness (dsh) out-of-tree plugins — tools, services, adapters, hooks, bundles. Use when writing a new plugin, adding a model tool, creating apply()/inject/Config, scaffolding a dsh.bundle package, or the user says /dsh-plugin-author, "写插件", "做个插件", "scaffold plugin", "加一个工具".
dsh-plugin-client
DeepSeek Harness browser-half plugins — dsh.client, client-modules, slots, settings cards, Conversation Nodes. Use when adding Web UI, slots.register, a settings tab, a chat node, or the user says /dsh-plugin-client, "浏览器插件", "写 UI 插件", "slot", "设置页卡片".
dsh-plugin-compose
Compose, install, and publish DeepSeek Harness plugin bundles — profile layers, cordis.patch.yml, dsh plugin add, --patch overlays, dump-config. Use when installing a plugin, writing a bundle patch, publishing to a profile, debugging layer order, or the user says /dsh-plugin-compose, "安装插件", "dsh plugin add"…