2672243194/dsh-read-url
A URL reader for DeepSeek Harness that fetches any URL - webpages (HTML), JSON APIs, RSS/Atom feeds - auto-detects the encoding (UTF-16 BOM, GBK/GB2312, UTF-8, Big5, Shift-JIS) and returns token-efficient compact text or structured Markdown from the clean main content, auto-joining paginated articles. Zero runtime dependencies, no API key and no server side; it plugs into the ctx.web seam and returns a paragraph-aligned truncation (default 6,000 chars) instead of the official web_fetch whole-page turndown.
dsh-read-url reads a URL into clean body text: it fetches any URL - webpages (HTML), plain text, Markdown, CSV, JSON and vendor +json APIs, and RSS 1.0/2.0 or Atom feeds including namespace prefixes, relative article links and xml:base - auto-detects the encoding (UTF-16 BOM, GBK/GB2312, UTF-8, Big5, Shift-JIS, with a mojibake fallback), extracts the main content and auto-joins paginated articles (default three pages), then returns token-efficient compact text or structured Markdown. Where the official web_fetch does a whole-page turndown conversion capped at 200,000 characters, this returns only the cleaned body plus essential metadata with a paragraph-aligned 6,000-character default truncation, a 5-minute session cache, and a full browser UA so middlebox/degraded-page pitfalls are avoided.
Install
npx @deepseek-ai/dsh plugin --profile web add github:2672243194/dsh-read-urlCopied verbatim from the README's install block, which uses the npx @deepseek-ai/dsh form. A local-checkout form (npx @deepseek-ai/dsh plugin --profile web add ./dsh-read-url) is also documented.
Compatibility
Zero runtime dependencies (Node 20+ built-ins handle fetch/decode/extract), no API key, no server side. It is seam-first on ctx.web with a fallback included, registers ctx.effect cleanup, and uses a cooperative timeout via timeoutMs + exec.signal so the elapsed time stays hidden from the model. An optional enhancement auto-enables headless-Chromium rendering for SPA/JS-redirect shells when Playwright is installed, degrading with an install hint when it is not (so the core stays zero-dependency).
Details
- Repo: 2672243194/dsh-read-url
- Category: Tools & Capabilities
- Stars: 20
- Version: GitHub repo 2672243194/dsh-read-url (20★, MIT); npm package dsh-read-url at 1.9.2 (registry check 2026-10-12; registry repository field resolves back to this repo). The github: form pins to the repo itself.
Alternatives
1321928757/dsh-api-kit · 1321928757/dsh-mysql · 1624318455/dsh-plugin-tavily