Web Access
Paste the following prompt into your AI chat to install this skill:
Follow https://skillhub.cn/install/skillhub.md to install @user_3c6cb52e/web-access-png.
About this skill
Problem
When agents browse the web, the hard part is not searching; it is reliably obtaining first-hand content. Anti-scraping rules, login walls, lazy-loaded DOM, screenshots, video sampling, and missing URL parameters can make search snippets or naive fetching misleading. web-access turns web access into a workflow with process validation, letting the agent switch between search summaries, page extraction, and browser CDP operations.
How It Works
The skill starts by defining the success criterion, then chooses the best entry point. For known URLs and article text, it can use WebFetch, curl, or Jina preprocessing. For logged-in content, interaction-heavy pages, or platforms where static fetching is unreliable, it uses browser CDP. The CDP proxy connects to the user's existing Chromium-based browser and works in background tabs via /eval, /click, /scroll, and /screenshot. It can inspect the DOM, extract media URLs, capture video frames, and navigate like a human. Independent research tasks can be split to parallel subagents, while the main agent receives only summaries.
Boundaries
Use it for fact-checking, platform research, logged-in content access, and web automation probes; avoid it for trivial single-page queries or high-stakes operations that require human judgment. CDP use should only close tabs created by the agent, not user tabs. Rapid batch navigation may trigger anti-bot controls. Site-pattern files are hints, not guarantees; fall back to generic behavior when they fail and update verified facts.
Use Cases
- When researching vendor announcements, access the official page directly and extract key release details instead of relying on conflicting snippets.
- After logging into an internal system, click exports, paginate, and extract report links in a background tab.
- When verifying policy terms, locate the issuing agency's original page and extract the applicable conditions.
- When product updates explain features in a video, seek to a timestamp and screenshot the relevant frame.
Best For
- Engineers who need agents to extract logged-in page content when static fetching fails.
- Analysts verifying claims and distinguishing official sources from media rewrites.
- Developers building agent web workflows and dispatching parallel research to subagents.
- Technical staff studying anti-bot site behavior and maintaining site-pattern notes.
Related Skills
A free matrix publishing assistant for Skill developers, guiding distribution to Tencent SkillHub, XiaLiao, XiaYou SkillHub, GitHub, and ClawHub with create, update, and delete safeguards.
A guide for writing Seedance 2.0 video prompts, covering @ references, camera moves, time segments, audio direction, and common pitfalls.
Collects daily conversations, task logs, and errors, then runs a nine-module review to surface blind spots, generate actions, and update memory.
Local long-term memory for AI agents that appends key facts and daily logs, supports cross-session context and search recall, and auto-archives old diaries to gzip files.