[
    {
        "role": "user",
        "content": "give%20some%20top%20open%20claw%20skills%20ranked%20highly%20for%20popularity%20or%20usefulness%20or%20novelty"
    },
    {
        "role": "assistant",
        "content": "Here are some of the top \u201cOpenClaw\u201d skills (in the sense of modules/extensions within the OpenClaw/ClawHub ecosystem) that are currently highly ranked\u2014based on popularity, usefulness, or novelty. I've grouped them by use case to help you pick what might fit best for you.\n\n---\n\n## \ud83d\udd0d Top OpenClaw Skills by Use Case\n\nThese are skill sets that are standout either because they get used a lot, solve real problems, or introduce new capabilities.\n\n### 1. Research & Knowledge Management\n\n- **Deep Research** \u2014 A standout across many workflows. It\u2019s multi-source, generates synthesized outputs with citations, and shows up in 5 out of 8 curated \u201cskill stacks.\u201d ([openclaw-skillshub.com](https://openclaw-skillshub.com/articles/best-openclaw-skills-by-use-case?utm_source=openai))  \n- **Notion Sync** \u2014 Keeps your Notion workspace aligned with other inputs and outputs; it\u2019s the core of the knowledge management stack. ([openclaw-skillshub.com](https://openclaw-skillshub.com/articles/best-openclaw-skills-by-use-case?utm_source=openai))  \n\n### 2. Development & Coding\n\n- **Code Reviewer** \u2014 Automates code review with checks for security, style, etc. A critical module in development-oriented stacks. ([openclaw-skillshub.com](https://openclaw-skillshub.com/articles/best-openclaw-skills-by-use-case?utm_source=openai))  \n- **Codebase-Agent** \u2014 Gives OpenClaw deep insight into your codebase: cross-file relations, architecture understanding, bug tracing, etc. ([openclaw101.vip](https://openclaw101.vip/blog/best-openclaw-skills-2026?utm_source=openai))  \n- **GitHub-Skill** \u2014 Use GitHub via natural language: PRs, merging, issue management, CI checks, etc., all controlled through OpenClaw. ([openclaw101.vip](https://openclaw101.vip/blog/best-openclaw-skills-2026?utm_source=openai))  \n\n### 3. Email & Automation\n\n- **Gmail Connector** \u2014 Connects your Gmail to OpenClaw; automates sorting, management, and even response suggestions. Big ROI for those swamped with email. ([openclaw-skillshub.com](https://openclaw-skillshub.com/articles/best-openclaw-skills-by-use-case?utm_source=openai))  \n- **Browser Pilot** \u2014 Allows automating web tasks without writing code, perfect for scraping, filling forms, or routine browser workflows. ([openclaw-skillshub.com](https://openclaw-skillshub.com/articles/best-openclaw-skills-by-use-case?utm_source=openai))  \n\n### 4. Productivity & Ops\n\n- **Morning Briefing Agent** \u2014 Sends structured morning summaries including tasks, deadlines, and suggested focus areas; improves daily planning and consistency. ([open-claw.sh](https://www.open-claw.sh/blog/best-openclaw-skills-2026?utm_source=openai))  \n- **Document Summarizer** \u2014 Jump in and extract the key points from long docs instead of reading the whole thing. Popular for obvious reasons. ([open-claw.sh](https://www.open-claw.sh/blog/best-openclaw-skills-2026?utm_source=openai))  \n\n### 5. Novel / Evolving Skills\n\n- **SkillClaw** \u2014 Rather than being a single skill, this is a framework for evolving skills collectively over time: it learns from user behavior, aggregates patterns, then improves or adds to the skill set. ([arxiv.org](https://arxiv.org/abs/2604.08377?utm_source=openai))  \n- **ABot-Claw** \u2014 Focuses on robotic or embodied agents that act in real-world environments with long-term memory and feedback loops. Useful for anyone pushing beyond purely digital agents. ([arxiv.org](https://arxiv.org/abs/2604.10096?utm_source=openai))  \n\n---\n\n## \ud83c\udfc6 Top Skills by Popularity\n\nThese are skills that are widely used and highly rated:\n\n- **self-improving-agent** \u2014 Has massive download counts and rising activity. ([topclawhubskills.com](https://www.topclawhubskills.com/?utm_source=openai))  \n- **Polymarket** \u2014 High user interest; one of the top trending modules. ([topclawhubskills.com](https://www.topclawhubskills.com/?utm_source=openai))  \n- **Skill Vetter** \u2014 Helps evaluate other skills or agent performance; often shows up among the most-used. ([topclawhubskills.com](https://www.topclawhubskills.com/?utm_source=openai))  \n\n---\n\n## \u2705 Best Picks Depending on What You're After\n\n| If you want\u2026 | Try installing |\n|-------------|----------------|\n| Better dev speed & code quality | *Code Reviewer*, *Codebase-Agent* |\n| Smarter research & content | *Deep Research*, *Notion Sync* |\n| Email & time management relief | *Gmail Connector*, *Morning Briefing Agent* |\n| Scaling agents, automation, robotics | *SkillClaw*, *ABot-Claw* |\n\n---\n\nIf you'd like, I can pull up the **top 5 skills** in any of these categories *right now* with install commands and steps\u2014would you prefer that?"
    },
    {
        "role": "user",
        "content": "the%20Browser%20Pilot%20sounds%20particularly%20interesting.%20what%20well-known%20websites%20does%20it%20work%20well%20with%3F%20which%20ones%20are%20more%20problematic%3F"
    },
    {
        "role": "assistant",
        "content": "Here\u2019s what I found out about *Browser Pilot* (aka the *Browser Automation / Browser* skill in OpenClaw) in terms of what works well, and what often causes trouble when automating websites. I\u2019ve grouped them into contexts of strong success vs problematic issues, with real-world examples and tips.\n\n---\n\n## \u2705 Websites & Scenarios That Work Well\n\nThese types of sites usually present fewer obstacles when using OpenClaw\u2019s browser skill:\n\n- **E-commerce product pages**  \n  Sites like Amazon or smaller online stores \u2014 grabbing prices, product titles, images, stock info, etc. \u2014 tend to work smoothly because they\u2019re relatively straightforward, JS-heavy but stable in structure. Users often use Browser Automation to \u201cGet the prices from these 5 Amazon product URLs.\u201d ([open-clawai.com](https://www.open-clawai.com/en/features/browser?utm_source=openai))\n\n- **Forms & applications with standard authentication**  \n  Filling out things like job application forms, surveys, or login + navigate flows on common dashboards. The system can \u201cfill forms, extract data, interact with any website\u201d via Playwright-based automation. ([open-clawai.com](https://www.open-clawai.com/en/features/browser?utm_source=openai))\n\n- **Monitoring & alerts workflows**  \n  Price drop alerts, shipping status, or tracking changes on a product page or dashboard \u2013 all good fits. The browser skill supports screenshot captures, structured data scraping, and \u201cwatching pages for changes\u201d features. ([open-clawai.com](https://www.open-clawai.com/en/features/browser?utm_source=openai))\n\n- **JS-heavy sites with dynamic content**  \n  So long as the site works in a regular Chromium browser (or Chrome-based browsers), Browser Automation tends to handle it fine. It supports full JS loading, login credentials, dynamic elements, etc. ([openclaw-browser.com](https://openclaw-browser.com/?utm_source=openai))\n\n---\n\n## \u26a0\ufe0f Websites or Cases That Are More Problematic\n\nThese tend to trip up OpenClaw\u2019s browser skill \u2014 not because they are impossible, but because of security, detection, or structural barriers.\n\n| Problem Type | What tends to happen | Examples / User-Reports |\n|--------------|----------------------|--------------------------|\n| **Sites that block automation / headless browsers** | Pages detect headless or scripted browsers and refuse to load, throw CAPTCHAs, or show \u201cThis browser isn\u2019t supported.\u201d | Users report running into \u201cmediocre results\u201d because of CAPTCHAs; Google sites often reject sign-ins via automation marked \u201cunsecure\u201d ([reddit.com](https://www.reddit.com/r/openclaw/comments/1rx76d1/why_your_openclaws_browser_keeps_getting_captchas/?utm_source=openai)) |\n| **High-security login/auth bastions** | Google, financial, or health portals which require 2FA, or detect unusual IPs/devices. They may block login, show extra verification, or prevent headless tools entirely. | \u201cMy OpenClaw is refusing to sign into websites\u2026 \u2018This browser or app may be unsecure\u2019\u2026\u201d when trying with Google login flows. ([reddit.com](https://www.reddit.com/r/openclaw/comments/1rv3fpm/my_openclaw_is_refusing_to_sign_into_websites/?utm_source=openai)) |\n| **Homespun intranets or internal dashboards** | Sites that rely on non-\u516c\u5f00 APIs, complex frames, heavy real-time push or sockets, or internal DNS. May not be reachable or block outside agents. | Users report \u2018Navigation failed\u2019 due to SSRF/network policy restrictions or the site being blocked because agent is out-of-network. ([reddit.com](https://www.reddit.com/r/openclaw/comments/1rsxraf/anyone_struggling_with_openclaw_browser/?utm_source=openai)) |\n| **Sites that change structure often** | When selectors change, CSS classes, or dynamic IDs, scraping or form filling scripts break. Sites using frequent updates or dynamically generated content can require constant fixes. |\n| **CAPTCHA/anti-bot protections** | Frequent issue. Even with real browser engines, many sites employ CAPTCHAs, device fingerprinting, or block automation. Leads to breaks or requests for manual input. ([reddit.com](https://www.reddit.com/r/openclaw/comments/1rx76d1/why_your_openclaws_browser_keeps_getting_captchas/?utm_source=openai)) |\n\n---\n\n## \ud83d\udd27 What the Docs Say: Capabilities & Limitations\n\nHere are official notes on what the browser skill provides, and under what constraints:\n\n- It supports *managed Chromium-based* browser profiles or attaching to your existing Chrome/Brave/Edge profile via Chrome DevTools MCP. So if you want to preserve existing logins/cookies, that path works. ([docs.openclaw.ai](https://docs.openclaw.ai/tools/browser?utm_source=openai))  \n- Many tasks\u2014click, type, navigate, screenshot, evaluate JavaScript\u2014are supported. ([docs.openclaw.ai](https://docs.openclaw.ai/tools/browser?utm_source=openai))  \n- There are some limitations: longer timeouts for JS-heavy pages, delays, or blocking behavior may require configuring the browser settings (_.timeout_, user agent, disabling \u201cautomated\u201d flags) for better stealth. ([crewclaw.com](https://www.crewclaw.com/blog/openclaw-browser-troubleshooting?utm_source=openai))\n\n---\n\n## \ud83d\udca1 Tips If You Hit Trouble\n\nIf you try to automate a site and it fails, here are adjustments that often help:\n\n- Use **existing-session profiles** so you inherit cookies/logins from a human-used browser. Avoid headless mode if possible for sites that detect headless. ([docs.openclaw.ai](https://docs.openclaw.ai/tools/browser?utm_source=openai))  \n- Adjust the user agent and disable flags like `--disable-blink-features=AutomationControlled`. Helps evade detection. ([crewclaw.com](https://www.crewclaw.com/blog/openclaw-browser-troubleshooting?utm_source=openai))  \n- Extend navigation and load timeouts. JS-heavy or slow pages often exceed default limits. ([crewclaw.com](https://www.crewclaw.com/blog/openclaw-browser-troubleshooting?utm_source=openai))  \n- Use screenshot / snapshot tools for debugging what the agent actually sees on page load. Misleading DOM or hidden elements are common.  \n- Be aware of legal/security issues: scraping or automating interactions with certain sites can violate terms of service, or trigger security controls.  \n\n---\n\nIf you tell me **which specific site(s)** you\u2019re thinking about (Amazon, Gmail, a bank portal, etc.), I can test or check whether Browser Pilot works well there (and whether people report issues). Want me to check for your sites of interest?"
    }
]