ai-search-browser-use-main_skill

This skill enables AI-driven web research using Chrome CDP for authenticated Gemini and Qwen queries with consolidated, cited results.
  • Python

2

GitHub Stars

2

Bundled Files

2 months ago

Catalog Refreshed

4 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill jwcodewrote/agent_skills_plugin --skill ai-search-browser-use-main

  • ai_query.py2.8 KB
  • SKILL.md12.3 KB

Overview

This skill enables reliable AI-assisted web research by automating a real Chrome browser using Chrome DevTools Protocol (CDP), with a browser-use fallback. It runs Gemini and Qwen queries in a controlled browser session, consolidates their responses, and returns a single synthesized answer with explicit citations and a note about which automation method was used.

How this skill works

The skill launches or connects to a Chrome instance running with remote debugging and a cloned user profile so logged-in sessions for Gemini and Qwen are available. It finds the AI pages via the CDP HTTP endpoint, injects input into the chat editors, triggers sends, waits for responses, extracts the reply text, and runs the same flow against both engines. Results are merged into a concise synthesis that surfaces consensus, disagreements, and source citations; if CDP is unavailable, the skill falls back to browser-use automation.

When to use it

  • When you need up-to-date web research using Gemini and Qwen in a real, logged-in browser session.
  • When answers should include explicit, verifiable citations from AI responses or original sources.
  • When authenticated access (your Chrome profile) is required to reach private or gated AI interfaces.
  • When you want side-by-side outputs from multiple AI engines consolidated into one answer.
  • When automation reliability is important and a robust fallback path is needed.

Best practices

  • Always run CDP with a non-default user-data-dir cloned from a logged-in profile to preserve authentication.
  • Verify Chrome is listening on the remote-debugging port before sending queries (curl http://localhost:9222/json).
  • Allow longer wait times (30–60s) for complex prompts or long responses to avoid partial extractions.
  • Close AI tabs or the CDP Chrome instance after the job to free port 9222 and avoid stale WebSocket URLs.
  • If fallback browser-use is used, clearly note it in the output and do not claim CDP automation.

Example use cases

  • Comparative expert summary: ask both Gemini and Qwen for pros/cons of a technical choice and synthesize the consensus.
  • Citation-backed fact-checking: query claims, extract sources, and provide direct links to primary sources.
  • Gated content retrieval: access responses that require an authenticated profile (subscription or single-sign-on).
  • Rapid literature scan: ask both engines for summaries of recent research and consolidate results with citations.

FAQ

Close the previous CDP Chrome instance (kill by command matching remote-debugging-port=9222) or choose a different port and relaunch with the same flags.

How do I know which automation method was used?

The skill explicitly reports whether CDP or the browser-use fallback was used in the final output.

Built by
VeilStrat
AI signals for GTM teams
© 2026 VeilStrat. All rights reserved.All systems operational
ai-search-browser-use-main skill by jwcodewrote/agent_skills_plugin | VeilStrat