deepseek-harness
deepseek-ai
DeepSeek Harness: Everything is a Plugin.
PROJECT TOPICS
INSTALL REFERENCE
dsh plugin --profile web add github:lywusichen/dsh-jmcomic
该命令指向仓库当前默认分支;尚无绑定当前 commit 的完整验证结果。
PROJECT README
English | 简体中文
A DeepSeek Harness plugin. Adds a comic button to the sidebar: search and download from JMComic (18comic), browse a local library, and read inside a window without leaving DSH.
The plugin bundles the download engine and the reader frontend, so it works without installing the jmcomic PyPI package.
Tested on DSH 0.1.5-rc.1 (web profile). Also runs on 0.1.2-rc.1 — those two
releases are what dsh.engines.dsh in package.json declares, so a host or
plugin manager can read the same range without parsing this file.
Uses the sidebar.footer.action and settings.section slots, the host's
workspaces.pickDirectory, and webServer.register for its local routes. The
DSH core is not modified.
Add the plugin:
dsh plugin --profile web add github:lywusichen/dsh-jmcomic
Restart DSH.
A comic button appears in the sidebar foot, above the skills button. On first use it asks for a local library folder; set the Python path in Settings → jmcomic Settings if it is not detected.
downloading banner appears at the top while a download runs.


The plugin ships four things under vendor/, so a machine with only DSH and
Python can run it:
jmcomic 2.7.3 source — the download enginecommonX — the engine's support packagejmcomic-ai — helper scriptsMissing binary dependencies degrade instead of failing:
| Missing | Behavior |
|---|---|
| curl_cffi | falls back to requests |
| Pillow | no image decoding; files download but cannot be previewed |
| pycryptodome | prompts to install it |
npm install # esbuild
npm run build # generates lib/client.js and lib/server.cjs
Commit both artifacts. Git-hosted installs consume them directly.
~/.dsh/plugins/dsh-jmcomic/settings.json (Windows)Bd/{Atitle}/第{Pindex}話The download engine, the reader frontend, and some helper scripts come from
hect0x7's projects, vendored under vendor/:
Copyright belongs to their respective authors.
CLASSIFICATION EVIDENCE
系统优先读取 GitHub Topics,再与站内分类词典和词根规则比对。当前命中: 无有效分类标签。