Features
Read 1–3 public HTTPS pages you supply, extract static text, source dates and excerpts, then use your connected model to write cited Markdown notes. Useful for comparing articles or collecting material on a topic.
Usage examples
After installation, send any example below to your AI. Include the files or URLs it refers to.
Read the two public pages I provide, compare their main points, and include source links and dates.
Collect installation requirements from these three pages into research notes with quotations.
Summarize my supplied web sources in Chinese and save Markdown notes and source records.
Advanced usage: run in a terminal
These commands are for manual execution in the terminal on the computer or server running OpenClaw. For everyday use, send one of the chat examples above.
Replace these placeholders with actual paths before running:
SKILL_DIRECTORY: the installed Skill folder.WORKSPACE: the agent workspace path.EVIDENCE_FILE: the evidence-file path returned by the previous step.
Create urls.json in the workspace with an array of page URLs, then collect:
node "SKILL_DIRECTORY/scripts/web-notes.mjs" collect --workspace "WORKSPACE" --urls "WORKSPACE/urls.json" --language en
Have OpenClaw write a draft from the returned evidence, then save:
node "SKILL_DIRECTORY/scripts/web-notes.mjs" save --workspace "WORKSPACE" --evidence "EVIDENCE_FILE"
Replace placeholders with actual paths. Finished notes go into the workspace's web-notes/ folder.