answer-me-with-html: A DeepSeek Harness Skill That Answers Hard Questions With One HTML Page
qingyuna/answer-me-with-html
An agent skill for DeepSeek Harness that answers hard questions with a single readable HTML page: the model writes only a Markdown draft while a bundled CLI renders diagrams and CSS, using about 1/8 of the output tokens of hand-written HTML.
answer-me-with-html is an answer-rendering skill for DeepSeek Harness that turns a wall of text into a single well-formatted, offline-readable HTML page. It targets the fact that models hand-writing HTML is slow and token-hungry: the model writes only a short Markdown draft, while the SVG diagrams, CSS and page tags are produced by the CLI bundled with the skill. Once enabled you just ask questions as usual, with no extra setup, maintained by QingYunA under the MIT license.
How to Install
dsh plugin --profile web add github:qingyuna/answer-me-with-html- Category
- Tools & Capabilities
- Platform
- DSH Plugin
- Author
- qingyuna
- Distribution
- Plugin
answer-me-with-html Key Features
answer-me-with-html Repo Summary
What Does It Do?
Answer me with HTML is a DSH plugin for DeepSeek Harness that answers hard questions with a single offline HTML page instead of a wall of text. It solves the problem of models hand-writing HTML, which is slow and token-hungry: the model writes only a short Markdown draft, while the bundled CLI generates the SVG diagrams, CSS and HTML tags. Maintained by QingYunA under the MIT license, with its site at answer-me-with-html.com, the repository has over 3,000 stars and was last updated in October 2026.
Core Features
- One-page HTML answers: turns hard questions such as the TCP three-way handshake, how a repo's modules fit together, or Redis versus Memcached into a single readable, offline file.
- Far fewer tokens: the model outputs only a Markdown draft, about 612 tokens on average, roughly one eighth of hand-written HTML, and about 2.6x faster.
- Automatic diagrams and styling: SVG coordinates and paths, CSS and HTML tags are all produced by the bundled CLI, not by the model.
- Follows the ASD-STE100 plain technical English standard and supports explainer-video output, which saves about 18x in output tokens in a small test.
- Always-on mode: once enabled, the agent attaches a page to every conclusion, not just the hard ones.
How to Use This Plugin?
After enabling it in DSH, just ask questions the way you always do, for example explaining the TCP three-way handshake, mapping out how the modules in a repository fit together, or comparing Redis with Memcached for caching. The agent writes a short Markdown draft and hands it to the CLI that ships with the skill; about 50 ms later you have a finished page. No setup is needed after install, and if you want a page for every conclusion you can add one rule to your rules file to turn on always-on mode.
How to Troubleshoot This Plugin?
The plugin requires Node.js 20 or newer, and there is no separate dependency install step because the CLI is bundled inside the skill. If page generation fails, first check that your local Node version meets the requirement, then repair the setup following the official instructions; if your agent cannot open links, use the agent-oriented install instructions or the command-line install path instead.
This page is an independent rewrite of the plugin's official README — for authoritative documentation and the latest changes, refer to the source: qingyuna/answer-me-with-html. The plugin is third-party code that runs on your machine once installed; inclusion does not imply endorsement — please review the source before installing.