[일:] 2025년 02월 25일

로컬에서 구동하는 “딥 리서치” browser use

📌 Browser-Use란?

browser-use의 Web UI는 자동화 및 원격 브라우저 제어를 위한 인터페이스입니다. 이는 다양한 환경에서 웹 스크래핑, UI 자동화, API 호출을 더 쉽게 수행할 수 있도록 돕는 도구입니다.

🔹 1. 주요 기능

✅ 웹 브라우저 원격 제어

  • browser-use는 Puppeteer 및 Playwright 같은 헤드리스 브라우저 엔진을 활용하여 UI 자동화 및 원격 브라우징을 지원합니다.
  • Web UI를 통해 브라우저 세션을 관리하고, 직접 조작하거나, 자동화 스크립트를 실행할 수 있습니다.

✅ 웹 자동화 및 데이터 스크래핑

  • 특정 웹사이트에서 데이터를 자동으로 가져올 수 있도록 설정할 수 있습니다.
  • browser-use Web UI를 통해 XPath 및 CSS Selector를 사용하여 데이터를 추출할 수 있습니다.

✅ 멀티 인스턴스 지원

  • 여러 개의 브라우저 세션을 실행할 수 있으며, 이를 Web UI에서 관리할 수 있습니다.
  • 각 인스턴스는 독립적인 쿠키 및 세션을 가질 수 있어, 여러 계정 로그인 또는 멀티탭 작업이 가능합니다.

✅ 스크립트 실행 기능

  • JavaScript 또는 Python 스크립트를 Web UI에서 실행하여 자동화된 작업을 설정할 수 있습니다.
  • API 호출을 통해 브라우저 제어가 가능하므로, 외부에서 특정 액션을 트리거할 수 있습니다.

✅ 프록시 및 사용자 지정 헤더 지원

  • browser-use는 다양한 프록시 설정을 지원하여 IP 우회가 가능합니다.
  • Web UI에서 User-Agent, Referer, Headers 등을 커스텀 설정하여 크롤링 탐지를 회피할 수 있습니다.

즉, 이걸로 무엇을 할 수 있냐면, 쿠팡에 접속해서 브라운 9시리즈 면도기중에 로켓배송으로 가장 저렴한걸 장바구니에 담아서 결제해! 라는 명령을 내리면, 수행한다는 거죠.
(물론 매번 모든걸 한번에 수행하진 못합니다, 이상한 에러가 날 수 있고, 로그인을 해줘야 할 수도 있고 해서, 조금의 코딩은 필요할 수 있습니다)

그리고 또 한가지, 딥 리서치를 시킬 수 있는 기능이 존재 합니다.

예를 들어 위와 같이 딥리서치를 해 보면

아래와 같은 로그를 볼 수 있습니다.

Extracted page content:
Title: – YouTube URL Source: https://www.youtube.com/watch?v=mudcagzsb9A
Markdown Content: – YouTube =============== Search Watch later Share Copy link
Info Shopping Tap to unmute 2x Image 3 If playback doesn’t begin
shortly, try restarting your device. Image 4 • Share Include playlist An
error occurred while retrieving sharing information. Please try again later.
Image 5 0:00 0:00 / 0:00•Live

  • • Scroll for details • NaN / NaN Back
    Search

INFO [src.agent.custom_agent] ✅ Task completed successfully
INFO [src.agent.custom_agent] Created GIF at agent_history.gif
INFO [controller] 🔗 Opened new tab with https://www.mdpi.com
INFO [controller] 🔄 Switched to tab 0
INFO [src.utils.deep_research] Extracted page content:
Title: https://www.google.com/search?q=LWIR%20vs%20ESWIR&udm=14 URL Source:
https://www.google.com/search?q=LWIR%20vs%20ESWIR&udm=14 Markdown Content:
https://www.google.com/search?q=LWIR%20vs%20ESWIR&udm=14 =============== * * *

  • * * About this page Our systems have detected unusual traffic from your
    computer network. This page checks to see if it’s really you sending the
    requests, and not a robot. Why did this happen? This page
    appears when Google automatically detects requests coming from your computer
    network which appear to be in violation of the Terms of Service. The block will expire
    shortly after those requests stop. In the meantime, solving the above CAPTCHA
    will let you continue to use our services. This traffic may have been sent by
    malicious software, a browser plug-in, or a script that sends automated
    requests. If you share your network connection, ask your administrator for
    help — a different computer using the same IP address may be responsible.
    Learn more Sometimes you
    may be asked to solve the CAPTCHA if you are using advanced terms that robots
    are known to use, or sending requests very quickly. IP address:
    2600:1900:0:2d04::2100 Time: 2025-02-25T08:24:04Z URL:
    https://www.google.com/search?q=LWIR%20vs%20ESWIR&udm=14

INFO [src.agent.custom_agent] ❌ Failed to complete task in maximum steps
INFO [src.agent.custom_agent] Created GIF at agent_history.gif
INFO [src.utils.deep_research] save query: MWIR camera specifications and uses at ./tmp/deep_research/ad7a689d-1305-4b5a-93b1-7c6b0d4cbac9\query_results\1-1.md
INFO [src.utils.deep_research] 🤯 Start Record Deep Thinking:
INFO [src.utils.deep_research]
Okay, so I need to figure out how to process the user’s query about SWIR, MWIR, LWIR, and ESWIR cameras. The user wants a detailed comparison between these types of cameras. Let me start by understanding each acronym.

First, SWIR stands for Short-Wave Infrared. These cameras operate in the 0.9 to 1.7 micrometer range. They’re often used in applications where some NIR wavelengths are blocked, like in glass or certain materials. I remember reading that SWIR is good for things like thermal imaging and can sometimes see through fog.

Next up is MWIR, which stands for Mid-Wave Infrared. This operates from 1.7 to 5 micrometers. MWIR cameras are typically used in military and surveillance because they work well in both day and night conditions. They can detect a wide range of thermal emissions, making them versatile.

Then there’s LWIR, or Long-Wave Infrared, covering 5 to 8 micrometers. These are often used for high-temperature objects like industrial furnaces or metal processing. They have better resolution at longer distances but might not be as effective in detecting cooler objects.

ESWIR is Extended Short-Wave Infrared, operating from 0.9 to 1.7 micrometers, similar to SWIR but maybe optimized differently. I’m a bit fuzzy on the exact differences here; perhaps ESWIR offers better sensitivity or specific spectral bands?

Looking at the current search results provided, it’s mostly about the FLIR A6780 MWIR camera. The specs mention 1.0 to 5.0 micrometers, which fits the MWIR range. It uses InSb detectors and has features like motorized focus and Gigabit Ethernet connectivity. This suggests that MWIR cameras are designed for high thermal sensitivity and flexibility in various applications.

I should structure the response by defining each camera type, their spectral ranges, key features, typical use cases, advantages, and limitations. Including technical specs like detector types (like InSb for MWIR) adds depth. I also need to ensure that each section is concise but informative, avoiding redundancy since the user emphasized minimizing information loss.

I should make sure to highlight how each camera type differs in terms of resolution, temperature range, and application suitability. For example, SWIR might be better for certain industrial inspections where LWIR isn’t as effective due to shorter wavelengths capturing different thermal signatures.

Finally, I’ll organize all this into a JSON list with summaries and reasoning on how each piece fits into the overall report. This way, the user gets a clear comparison that can be easily referenced or used in further analysis.

INFO [src.utils.deep_research] 🤯 End Record Deep Thinking
INFO [src.utils.deep_research] 🤯 Start Record Deep Thinking:
INFO [src.utils.deep_research]
Okay, so I need to figure out the differences between SWIR, MWIR, LWIR, and ESWIR cameras. The user provided some previous information which includes details on each type except for MWIR, which they just searched about. Let me start by recalling what I know.

From the Previous Recorded Information, I remember that SWIR (Short-Wave Infrared) operates in 0.9 to 1.7 µm and is good for detecting thermal emissions through glass. It’s used in industrial inspections and surveillance. LWIR (Long-Wave Infrared) covers 5 to 8 µm, ideal for high-temperature applications like furnaces but struggles with cooler objects. ESWIR extends SWIR into NIR, making it better for advanced surveillance.

Now, the current search was about MWIR cameras. The previous info had a FLIR A6780 MWIR camera which operates from 1.0 to 5.0 µm. It uses a cooled InSb detector and has high frame rates. This suggests MWIR is used in industrial and military settings, good for detailed thermal analysis.

I need to structure the JSON output with each camera type as separate entries. Each should have their spectral range, key features, applications, advantages, and disadvantages. The thinking field should explain where this info fits in a report—like introduction or analysis sections.

Wait, I should make sure not to repeat information already present. For example, SWIR’s ability to penetrate glass was mentioned before, so I don’t need to include that again for MWIR unless it adds new context.

I also need to ensure each entry has the correct URL and title if available. Since the current search result didn’t yield useful info beyond the YouTube video (which seems unhelpful), most entries will have “unknown” as source.

Let me outline each camera type:

  1. SWIR: 0.9-1.7 µm, glass penetration, industrial/surveillance.
  2. MWIR: 1.0-5.0 µm, cooled detectors, high frame rates, industrial/military.
  3. LWIR: 5-8 µm, high temp apps, poor for cool objects.
  4. ESWIR: Extends SWIR to NIR, advanced surveillance.

Each should have a summary with key points and thinking on placement in the report.

I think I’ve covered all necessary details without redundancy. Now, structuring each as JSON entries with appropriate titles and URLs where possible.

INFO [src.utils.deep_research] 🤯 End Record Deep Thinking
INFO [src.utils.deep_research] save query: LWIR camera technology and applications at ./tmp/deep_research/ad7a689d-1305-4b5a-93b1-7c6b0d4cbac9\query_results\1-2.md
INFO [src.utils.deep_research] 🤯 Start Record Deep Thinking:
INFO [src.utils.deep_research]
Okay, so I need to figure out the differences between SWIR, MWIR, LWIR, and ESWIR cameras. The user provided some previous information and a new search result about LWIR cameras. Let me start by recalling what I know from the previous data.

From before, I remember that SWIR (Short-Wave Infrared) operates in 0.9 to 1.7 µm. They can detect thermal emissions through glass and are used in industrial inspections and surveillance. MWIR (Mid-Wave Infrared) is from 1.0 to 5.0 µm, uses cooled detectors like InSb, and is good for high frame rates. It’s used in military and manufacturing. LWIR (Long-Wave Infrared) covers 5 to 8 µm, ideal for high-temperature objects and longer distances but struggles with cooler ones. ESWIR extends SWIR into NIR, up to 2.6 µm, making them better for advanced surveillance.

Now, looking at the new search result about LWIR cameras from IDTechEx. It mentions that LWIR is used in automotive for ADAS and autonomous driving because they can detect pedestrians and objects in low-light conditions. They’re passive systems using microbolometers, which are reliable but expensive. The report also talks about their future growth due to regulations like NHTSA’s requirements.

I need to extract the key points from this new result without duplicating previous info. The summary should include spectral range, technology used (microbolometers), applications in automotive for safety, and market trends regarding cost and adoption.

Thinking about how this fits into a report: it would go under the analysis section, comparing LWIR with other types. It highlights their niche but growing role in ADAS and autonomous vehicles, which ties into real-world applications and technological advancements.

INFO [src.utils.deep_research] 🤯 End Record Deep Thinking
INFO [src.utils.deep_research] 🤯 Start Record Deep Thinking:
INFO [src.utils.deep_research]
Okay, so I need to figure out the differences between SWIR, MWIR, LWIR, and ESWIR cameras. The user provided some previous information and a new search result about LWIR cameras. Let me start by recalling what I know from the previous data.

From before, I remember that SWIR (Short-Wave Infrared) operates in 0.9 to 1.7 µm. They can detect thermal emissions through glass, which is useful for surveillance and industrial inspections. ESWIR extends this range into NIR, making them better for detecting a wider range of thermal emissions.

MWIR (Mid-Wave Infrared) covers 1.0 to 5.0 µm. The FLIR A6780 example had a cooled InSb detector and high frame rates, used in industrial and military settings. LWIR (Long-Wave Infrared) operates from 5 to 8 µm, ideal for high-temperature objects like furnaces. They have better resolution at longer distances but struggle with cooler objects.

Now, looking at the new search result about LWIR cameras: it’s part of a report by IDTechEx discussing their use in automotive applications. LWIR is used for detecting pedestrians and other road users in low-light conditions because they can pick up heat signatures effectively. The market is growing due to regulations like NHTSA’s mandates for advanced AEB systems.

I need to extract the key points from this new information without duplicating what’s already known. The previous data didn’t cover automotive applications, so that’s a new angle. Also, the report mentions how LWIR will be crucial for autonomous vehicles and robotaxis because of their reliability in low-visibility conditions.

So, I should summarize the new info about LWIR’s role in ADAS, autonomous driving, and regulatory compliance. This adds context to their application beyond industrial uses. It also touches on market growth projections, which is valuable for understanding their future significance.

I need to make sure the summary includes technical specs like spectral range, applications (especially automotive), and how they compare to other camera types in terms of use cases and effectiveness. The thinking part should link this info to sections like analysis or application-specific uses in a report.

INFO [src.utils.deep_research] 🤯 End Record Deep Thinking
INFO [src.utils.deep_research] 🤯 Start Record Deep Thinking:
INFO [src.utils.deep_research]
Okay, so I need to figure out the differences between SWIR, MWIR, LWIR, and ESWIR cameras. The user provided some previous information and a new search result about LWIR cameras. Let me start by recalling what I know from the previous data.

From before, I remember that SWIR (Short-Wave Infrared) operates in 0.9 to 1.7 µm. They can see through glass, which is useful for industrial inspections and surveillance. MWIR (Mid-Wave Infrared) covers 1.0 to 5.0 µm, used in military and manufacturing with cooled detectors. LWIR (Long-Wave Infrared) works from 5 to 8 µm, ideal for high-temperature objects and automotive applications like ADAS. ESWIR extends SWIR into NIR, up to 2.6 µm, for advanced surveillance.

Now, looking at the new search result about LWIR technology and applications. It’s a detailed report from IDTechEx. The content mentions that LWIR cameras operate in 5 to 8 µm, suitable for detecting high-temperature objects and performing well in low-light conditions. They’re increasingly used in automotive for ADAS and autonomous driving because they can detect pedestrians effectively.

The report also talks about market growth driven by regulations like NHTSA’s mandates for advanced AEB. It expects significant adoption in higher SAE level vehicles. Technologically, LWIR uses microbolometers and has potential for sensor fusion with other systems like LiDAR and radar. However, it’s currently limited to high-end vehicles but is expected to become more mainstream as costs come down.

So, how does this new info fit into what I already have? Previously, I noted that LWIR was used in industrial furnaces and had limitations with cooler objects. Now, the IDTechEx report adds automotive applications, regulatory influences, and technological advancements like sensor fusion. This provides a more comprehensive understanding of LWIR’s role beyond just industrial uses.

I should make sure not to repeat information already captured. The new data emphasizes automotive applications, which wasn’t as detailed before. It also introduces market forecasts and regulatory impacts, which are important for a thorough comparison.

In terms of structure, the summary should highlight the spectral range, key features (like microbolometers), applications (especially in ADAS and autonomous driving), and mention the influence of regulations. The thinking part should note that this info is crucial for understanding automotive uses and how LWIR compares to other camera types in safety technologies.

I need to ensure that the summary includes key data points like spectral range, detector type, frame rates if available, and applications. Also, since the source URL and title are provided, I’ll include those accurately.

INFO [src.utils.deep_research] 🤯 End Record Deep Thinking
INFO [src.utils.deep_research] 🤯 Start Record Deep Thinking:
INFO [src.utils.deep_research]
Okay, so I need to figure out the differences between SWIR, MWIR, LWIR, and ESWIR cameras. The user provided some previous information and current search results, so I should use those to create a JSON list with summaries and thinking.

First, looking at the Previous Recorded Information, there are entries about each camera type. For example, SWIR operates in 0.9-1.7 µm, good for detecting through glass. MWIR is from 1.0-5.0 µm, used in industrial and military settings with high frame rates. LWIR covers 5-8 µm, ideal for high-temperature objects and automotive uses like ADAS. ESWIR extends SWIR into NIR, useful for advanced surveillance.

The current search results focus on LWIR technology and applications. It mentions that LWIR cameras are used in automotive for AEB and autonomous driving because they can detect pedestrians well in low-light conditions. They’re passive systems relying on heat signatures, which makes them reliable but limited to higher-end vehicles currently due to cost.

I need to extract the key points from these sources without redundancy. Each entry should have a URL, title, summary, and thinking about where it fits in a report.

For LWIR cameras, the IDTechEx report provides detailed info on their use in ADAS, autonomous driving, and regulatory compliance. It also talks about cost reductions making them more accessible. This is important for understanding their role compared to other camera types.

I should structure each JSON entry with these details, ensuring that key data like spectral ranges and applications are included. The thinking part should explain how each piece can be used in the report—like introduction, background, analysis sections.

So, I’ll create entries for each camera type based on the provided info, making sure to include URLs where available and “unknown” where sources aren’t specified. Each summary will capture essential details without losing important figures.

INFO [src.utils.deep_research] 🤯 End Record Deep Thinking
INFO [src.utils.deep_research] save query: Difference between SWIR and MWIR cameras at ./tmp/deep_research/ad7a689d-1305-4b5a-93b1-7c6b0d4cbac9\query_results\1-4.md
INFO [src.utils.deep_research] 🤯 Start Record Deep Thinking:
INFO [src.utils.deep_research]
Okay, so I need to figure out the differences between SWIR, MWIR, LWIR, and ESWIR cameras. The user provided some previous information, but the current search results are just a Google error message about too many requests. Let me start by recalling what I know from the Previous Recorded Information.

From there, I remember that SWIR (Short-Wave Infrared) operates in 0.9 to 1.7 µm and is good for detecting thermal emissions through glass. It’s used in industrial inspections and surveillance. MWIR (Mid-Wave Infrared) covers 1.0 to 5.0 µm, uses cooled detectors like InSb, and is used in industrial, military, and R&D applications with high frame rates.

LWIR (Long-Wave Infrared) works from 5 to 8 µm, ideal for high-temperature objects and automotive ADAS. ESWIR extends SWIR into NIR up to 2.6 µm, enhancing detection capabilities for surveillance and research.

Since the current search didn’t yield new results due to the error, I can only rely on existing data. I should structure the JSON output with each camera type’s specs, applications, pros, cons, and where they fit in a report. Make sure not to repeat information already recorded and include key details like spectral ranges and detector types.

INFO [src.utils.deep_research] 🤯 End Record Deep Thinking
INFO [src.utils.deep_research] save query: Comparison of LWIR vs ESWIR cameras at ./tmp/deep_research/ad7a689d-1305-4b5a-93b1-7c6b0d4cbac9\query_results\1-5.md
INFO [src.utils.deep_research] 🤯 Start Record Deep Thinking:
INFO [src.utils.deep_research]
Okay, so I need to figure out the differences between SWIR, MWIR, LWIR, and ESWIR cameras. The user provided a detailed search result from Axiom Optics about these types of infrared cameras. Let me start by understanding each camera type based on the information given.

First, SWIR (Short-Wave Infrared) cameras operate in the range of 0.9 to 1.7 micrometers. They detect reflected light rather than emitted heat. This makes them useful for applications like detecting defects in fruits or electronics, counterfeit detection, and art restoration. The article mentions that Sony’s SenSWIR technology extends this range up to 2.2 micrometers, which is considered ESWIR (Extended SWIR). These cameras use materials like InGaAs, Ge, or InGaAsP for detection.

Next, MWIR (Middle-Wave Infrared) cameras work in the 3 to 5 micrometer range and detect emitted heat. They are effective in industrial settings for gas leak detection and in security for long-range surveillance. The article highlights that MWIR cameras have better penetration through atmospheric conditions compared to LWIR, making them suitable for outdoor use. They often require cooling to reduce noise and use detectors like InSb or HgCdTe.

Then there’s LWIR (Long-Wave Infrared) cameras, which operate from 8 to 14 micrometers. These are typically used in military applications because they can detect heat emitted by people and vehicles. They often use microbolometers and don’t require cooling as frequently as MWIR cameras. However, cooled versions offer higher resolution for specific needs.

ESWIR (Extended SWIR) extends the SWIR range beyond 1.7 micrometers up to around 2.2 micrometers. This allows them to detect a wider range of thermal emissions, useful in advanced surveillance and scientific research where longer wavelengths are necessary.

I should summarize each type’s key features, applications, and technologies used. It’s important to note the differences in their spectral ranges, whether they detect reflected light or emitted heat, and their typical use cases. Also, mentioning the materials and cooling requirements will help differentiate them further.

INFO [src.utils.deep_research] 🤯 End Record Deep Thinking
INFO [src.utils.deep_research] 🤯 Start Record Deep Thinking:
INFO [src.utils.deep_research]
Okay, so I need to figure out the differences between SWIR, MWIR, LWIR, and ESWIR cameras. The user has already provided some previous information, but now they want a comparison specifically between LWIR and ESWIR cameras.

Looking at the current search results, it seems like there was an error because Google returned a 429 Too Many Requests when trying to access the comparison. That means I can’t get direct information from that source. So, I have to rely on the previously recorded information and any other available data.

From the previous records, I know:

  • SWIR (Short-Wave Infrared) operates in the range of 0.9 to 1.7 µm. It’s good for detecting thermal emissions through materials like glass and is used in industrial inspections and surveillance.
  • MWIR (Mid-Wave Infrared) operates from 1.0 to 5.0 µm. The FLIR A6780 MWIR camera, for example, uses a cooled InSb detector and has high frame rates up to 4,130 Hz. It’s used in industrial, military, and manufacturing R&D.
  • LWIR (Long-Wave Infrared) operates from 5 to 8 µm. It’s ideal for high-temperature applications like industrial furnaces and is used in automotive ADAS and autonomous driving because it works well in low-light conditions.
  • ESWIR (Extended Short-Wave Infrared) extends the SWIR range into NIR, up to 2.6 µm. This allows better detection of a wider range of thermal emissions, making it suitable for advanced surveillance and scientific research.

Since I can’t access new information on LWIR vs ESWIR, I’ll have to compare them based on what’s already known.

LWIR operates at longer wavelengths (5-8 µm) and is better for high-temperature objects and low-light conditions. It’s used in automotive applications because it can detect pedestrians effectively.

ESWIR extends SWIR into NIR, offering a broader spectrum (0.9 to 2.6 µm). This makes it more versatile for detecting various thermal emissions, useful in surveillance and scientific research.

So, the key differences would be their spectral ranges, typical use cases, and applications. LWIR is better for high-temperature and automotive needs, while ESWIR offers extended capabilities for broader detection.

INFO [src.utils.deep_research] 🤯 End Record Deep Thinking
INFO [src.utils.deep_research] 🤯 Start Record Deep Thinking:
INFO [src.utils.deep_research]
Okay, so I need to figure out the differences between SWIR, MWIR, LWIR, and ESWIR cameras. The user provided some previous information, but it seems like the current search didn’t yield any useful results because Google blocked it with a CAPTCHA message. That’s not helpful.

Looking at the Previous Recorded Information, I see that there are summaries about each camera type. For example, SWIR operates in 0.9 to 1.7 µm and is good for detecting thermal emissions through glass. MWIR goes from 1.0 to 5.0 µm and is used in industrial and military settings. LWIR covers 5 to 8 µm and is ideal for high-temperature objects, while ESWIR extends SWIR into the NIR spectrum up to 2.6 µm.

Since the current search didn’t give any new data, I can only rely on what’s already been recorded. So, I should structure the JSON output based on that information without adding anything new from the blocked search results.

I need to make sure each entry is concise but includes key details like spectral ranges and applications. Also, I must avoid redundancy by checking if the same info is already present. For example, there are multiple entries about LWIR in automotive applications, so I should include that once.

The thinking field should explain how each piece of information can be used in a report, maybe in sections like introduction or analysis. Since all the data is from previous searches, the URLs and titles will either be known sources or marked as unknown if not attributed.

I’ll go through each entry in Previous Recorded Information, extract the necessary details, ensure they’re unique, and format them into JSON with appropriate thinking on their placement in a report.

INFO [src.utils.deep_research] 🤯 End Record Deep Thinking
<마치 추론 모델인양..열심히 추론하는 모습 ㄷㄷㄷㄷㄷㄷㄷ>

이런 식으로 결과물을 마크다운으로 생성합니다.

내용물을 보면 마크다운 문법으로 엄청난 양을 리서치를 해 놨네요;;
(참고로 3090 24g모델 한장으로 돌린거라 거진 30분 넘게 돌아간거 같습니다;;;;)
(1번 문서에는 출처가 정리되고 2~3번 문서에는 내용이 정리되어 있네요)

이렇게 딥리서치를 구동할 수 있는 장점이 있어서 퍼플렉시티나 챗gpt의 딥리서치 보다 조금 더 풍부한 리서치를 시킬 수 있는 장점이 있습니다.

이외 다양한 활용법으로는

🎯 1. 데이터 크롤링 (Web Scraping)

✔️ 예제: 뉴스 사이트에서 최신 기사 수집

  • 특정 뉴스 사이트에 접속하여 제목, 링크, 요약 등을 자동으로 가져올 수 있음.
  • CSS Selector 또는 XPath를 활용하여 원하는 요소를 추출 가능.

✅ 사용 예제

javascript복사편집const browser = await browserUse.launch();
const page = await browser.newPage();
await page.goto('https://news.ycombinator.com/');
const articles = await page.$$eval('.titleline a', links => links.map(link => link.innerText));
console.log(articles);
await browser.close();

📝 활용 예시

  • 특정 키워드가 포함된 뉴스만 자동으로 수집 후 Telegram 알림 전송
  • E-Commerce 사이트에서 특정 상품 가격 변동 감지

🎯 2. 자동 로그인 및 웹사이트 조작

✔️ 예제: Gmail 자동 로그인 및 메일 검색

  • ID/PW 입력 후 로그인하여 자동으로 특정 메일을 검색할 수 있음.
  • CAPTCHA가 있다면 AI Solver와 결합하여 자동 해결 가능.

✅ 사용 예제

javascript복사편집await page.type('#identifierId', 'your-email@gmail.com');
await page.click('#identifierNext');
await page.waitForSelector('input[type="password"]');
await page.type('input[type="password"]', 'your-password');
await page.click('#passwordNext');
await page.waitForNavigation();
console.log('로그인 완료');

📝 활용 예시

  • 특정 웹사이트 로그인 후 데이터 다운로드
  • 자동 로그인을 이용한 매크로 작업 (예: 온라인 강의 자동 출석)

🎯 3. 다중 계정 관리

✔️ 예제: 여러 계정으로 SNS 자동화

  • Twitter, Instagram, Facebook 등의 계정 여러 개를 유지하면서 자동화 가능.
  • 각 브라우저 세션이 독립적이므로 쿠키와 세션 충돌 없이 실행 가능.

✅ 사용 예제

javascript복사편집const browser1 = await browserUse.launch({ userDataDir: './user1' });
const page1 = await browser1.newPage();
await page1.goto('https://twitter.com/login');

const browser2 = await browserUse.launch({ userDataDir: './user2' });
const page2 = await browser2.newPage();
await page2.goto('https://twitter.com/login');

📝 활용 예시

  • 여러 개의 SNS 계정으로 자동 포스팅
  • E-commerce 다중 계정 운영 (예: 쿠팡, G마켓 자동 로그인 및 재고 확인)

🎯 4. UI 테스트 자동화

✔️ 예제: 웹사이트 UI 자동 테스트

  • 개발자가 만든 웹 애플리케이션이 정상적으로 동작하는지 확인할 수 있음.
  • 버튼 클릭, 입력, 페이지 이동 등 UI 테스트 시나리오를 작성하여 자동화 가능.

✅ 사용 예제

javascript복사편집await page.goto('https://your-website.com');
await page.click('#login-button');
await page.type('#username', 'testuser');
await page.type('#password', 'password123');
await page.click('#submit');
await page.waitForNavigation();
console.log('로그인 테스트 성공');

📝 활용 예시

  • 프론트엔드 개발자가 배포 전에 UI 자동 테스트 실행
  • 쇼핑몰 결제 프로세스 테스트 자동화

🎯 5. 특정 사이트 자동화 (예: 예약, 티켓팅)

✔️ 예제: 영화 예매 자동화

  • 특정 사이트에서 예매가 열리면 자동으로 좌석을 선택하고 결제 가능.

✅ 사용 예제

javascript복사편집await page.goto('https://cinema-site.com');
await page.click('#movie-select');
await page.select('#time-slot', '19:00');
await page.click('#seat-selection');
await page.click('#confirm-payment');
console.log('영화 예매 완료');

📝 활용 예시

  • 콘서트 티켓팅 자동화
  • 병원 예약 자동화

🎯 6. API 트래픽 분석 및 보안 테스트

✔️ 예제: 특정 사이트의 API 요청 자동 수집

  • 웹사이트에서 실행되는 모든 네트워크 요청을 기록하여 API 트래픽을 분석 가능.
  • 이를 활용하면 웹 보안 및 취약점 테스트(예: CSRF, XSS) 수행 가능.

✅ 사용 예제

javascript복사편집page.on('request', request => console.log('Request:', request.url()));
page.on('response', response => console.log('Response:', response.url()));
await page.goto('https://target-site.com');

📝 활용 예시

  • 특정 API가 요청하는 데이터 구조 분석 (예: 로그인 API, 결제 API)
  • 보안 테스터가 웹사이트의 취약점 점검

🎯 7. VPN 및 프록시 우회

✔️ 예제: 프록시 서버를 사용한 익명 브라우징

  • 특정 국가에서만 접속 가능한 사이트에 접근할 때 유용함.
  • 프록시를 설정하여 IP 차단 우회 가능.

✅ 사용 예제

javascript복사편집const browser = await browserUse.launch({
  args: ['--proxy-server=http://proxy-ip:port']
});

📝 활용 예시

  • 지역 제한이 걸린 사이트에 접속 (예: 해외 한정 스트리밍 서비스)
  • 크롤링 탐지를 우회하여 데이터를 수집

🎯 8. 채팅봇 및 고객 응대 자동화

✔️ 예제: 웹사이트 챗봇 자동 응답

  • 고객 문의를 자동 응답하는 챗봇을 구현할 수 있음.
  • Telegram, Discord 등의 API와 연동하여 자동 채팅 가능.

✅ 사용 예제

javascript복사편집await page.type('#chat-input', '안녕하세요! 무엇을 도와드릴까요?');
await page.click('#send-button');

📝 활용 예시

  • 웹사이트에 AI 챗봇 자동 응답 구현
  • Telegram과 연동하여 실시간 채팅 자동화

🎯 9. 광고 및 마케팅 자동화

✔️ 예제: 자동 댓글 및 좋아요

  • 특정 SNS 플랫폼에서 자동으로 댓글을 달거나 좋아요를 누를 수 있음.
  • 이를 활용하여 마케팅 자동화를 구축 가능.

✅ 사용 예제

javascript복사편집await page.goto('https://instagram.com/someprofile');
await page.click('.like-button');
await page.type('.comment-box', '좋은 게시물이네요!');
await page.click('.submit-comment');

📝 활용 예시

  • 인스타그램, 트위터 자동 댓글 및 좋아요
  • 상품 리뷰 자동화

위와 같은 것들을 할 수 있지요.

자 그럼 설치방법을 안내해 보겠습니다.

📌 Windows에서 browser-use 설치 방법 (단계별 가이드)

Windows 환경에서 browser-use를 설치하고 실행하는 방법을 단계별로 안내합니다.

🚀 1. 필수 요구 사항 확인

설치를 진행하기 전에 필요한 도구가 설치되어 있는지 확인하세요.

🔹 ✅ Node.js 설치

browser-use는 Node.js 기반으로 실행되므로, 먼저 Node.js를 설치해야 합니다.

  1. Node.js 공식 웹사이트에 접속합니다.
  2. LTS (Long-Term Support) 버전을 다운로드합니다.
  3. 설치 파일을 실행하고 기본 설정으로 설치를 완료합니다.
  4. 설치 후, 명령 프롬프트(CMD) 또는 PowerShell을 열어 다음 명령어를 입력하여 정상 설치 여부를 확인합니다.powershell복사편집node -v npm -v 예시 출력:scss복사편집v18.16.0 (Node.js 버전) 9.5.1 (npm 버전)

🔹 ✅ Git 설치 (필요한 경우)

browser-use 저장소를 클론(clone)하려면 Git이 필요합니다.

  1. Git 공식 웹사이트에서 최신 버전을 다운로드합니다.
  2. 설치 파일을 실행하고 기본 설정으로 설치합니다.
  3. 설치 후, 명령 프롬프트(CMD) 또는 PowerShell을 열어 다음 명령어를 입력하여 정상 설치 여부를 확인합니다.powershell복사편집git --version 예시 출력:nginx복사편집git version 2.41.0.windows.1

🚀 2. browser-use 다운로드 및 설치

이제 browser-use를 다운로드하고 실행할 준비를 합니다.

🔹 ✅ browser-use 클론 (Git 사용)

PowerShell 또는 CMD에서 실행

powershell복사편집git clone https://github.com/browser-use/browser-use.git
cd browser-use

위 과정이 끝나면 browser-use 폴더로 이동합니다.

🔹 ✅ 의존성 패키지 설치

powershell복사편집npm install
  • 이 명령어는 package.json에 정의된 모든 필수 모듈을 설치합니다.
  • 정상적으로 설치되면 node_modules 폴더가 생성됩니다.

🚀 3. browser-use 실행

이제 browser-use를 실행할 차례입니다.

🔹 ✅ 실행 명령어

powershell복사편집npm start
  • 실행이 성공하면 콘솔에 서버 실행 메시지가 출력됩니다.

✅ 기본 포트: http://localhost:3000
✅ Web UI 접속: 브라우저를 열고 http://localhost:3000으로 접속

🚀 4. 추가 설정 및 구성

🔹 ✅ 포트 변경

기본 포트(3000)을 변경하려면 .env 파일을 수정하거나 실행할 때 환경 변수를 설정할 수 있습니다.

1) .env 파일 수정

ini복사편집PORT=8080

2) 실행할 때 직접 포트 설정

powershell복사편집set PORT=8080 && npm start

🚀 5. browser-use를 백그라운드에서 실행

🔹 ✅ pm2를 사용하여 실행 (권장)

pm2는 Node.js 애플리케이션을 백그라운드에서 실행하는 도구입니다.

  1. pm2 설치powershell복사편집npm install -g pm2
  2. browser-use 실행powershell복사편집pm2 start npm --name "browser-use" -- start
  3. 실행 확인powershell복사편집pm2 list
  4. 시스템 시작 시 자동 실행 설정powershell복사편집pm2 startup pm2 save

🚀 6. 방화벽 및 네트워크 설정 (선택 사항)

만약 원격에서 browser-use에 접속하려면 Windows 방화벽에서 포트를 열어야 합니다.

🔹 ✅ Windows 방화벽에서 포트 열기

powershell복사편집netsh advfirewall firewall add rule name="browser-use" dir=in action=allow protocol=TCP localport=3000

✅ 사용자 컴퓨터에서 http://<IP>:3000 으로 접속 가능

🚀 7. 실행 및 테스트

설치가 끝났다면, 웹 브라우저에서 http://localhost:3000을 입력하여 browser-use Web UI가 정상적으로 작동하는지 확인하세요.

✅ 실행 후 정상적으로 로딩되면, browser-use를 활용하여 웹 자동화 및 크롤링을 진행할 수 있습니다.

🎯 정리

단계설명
1. 필수 요구 사항 확인Node.js 및 Git 설치
2. browser-use 다운로드Git으로 소스 코드 클론
3. 의존성 설치npm install 실행
4. 실행npm start 또는 pm2 사용
5. 포트 변경 (선택 사항).env 파일 수정 또는 실행 시 환경 변수 설정
6. 방화벽 설정 (선택 사항)Windows 방화벽에서 포트 열기
7. 테스트http://localhost:3000 접속


옵시디언(Obsidian) 이제 업무용으로도 무료 사용 가능!

그동안 개인 사용자는 무료였던 옵시디언, 그러나 업무용으로 사용하려면 라이선스 문제가 있었습니다.
그러나 20일 부터는 상업적 사용도 무료로 라이선스가 변경되었습니다.

자세한 내용은 아래 원문과 번역본을 참고해 주세요.

번역본

Obsidian 상업 라이선스가 선택 사항으로 변경되었습니다.

오늘부터 Obsidian의 상업 라이선스(Commercial License)가 선택 사항이 됩니다. 이제 누구나 업무용으로 무료로 Obsidian을 사용할 수 있습니다.
Obsidian이 귀하의 조직에 도움이 된다면, 여전히 상업 라이선스를 구매하여 개발을 지원할 수 있습니다.

변경되는 사항

✔ 아무것도 달라지지 않습니다.
✔ 계정이 필요하지 않습니다.
✔ 광고 없음, 추적 없음, 숨겨진 조건 없음.
✔ 데이터는 사용자의 완전한 통제 하에 있으며, 로컬에 Markdown 파일로 저장됩니다.
✔ 모든 기능을 제한 없이 무료로 사용할 수 있습니다.

왜 이러한 변화를 선택했을까요?

단순함(Simplicity) 때문입니다. 기존의 상업 라이선스 정책은 혼란을 초래하고 가격 구조를 불필요하게 복잡하게 만들었습니다. 또한, **Obsidian Manifesto(Obsidian 선언문)**에서 밝힌 바와 같이,

“우리는 모든 사람들이 명확하게 사고하고 효과적으로 아이디어를 정리할 수 있는 도구를 가져야 한다고 믿습니다.”
이러한 원칙을 더욱 충실히 따르기 위한 결정입니다.

10,000개 이상의 조직에서 Obsidian을 사용하고 있습니다.

Obsidian은 정부, 사이버 보안, 의료, 금융과 같은 보안이 중요한 환경에서 널리 사용됩니다.
Amazon과 Google을 포함한 세계 최대 규모의 조직에서도 수천 명의 직원이 매일 Obsidian을 사용하고 있습니다.
이 팀들은 보다 효과적으로 사고하고, 데이터의 완전한 소유권을 유지하기 위해 Obsidian을 신뢰합니다.

상업 라이선스가 선택 사항으로 변경됩니다.

기존에는 직원이 2명 이상인 기업이 Obsidian을 업무용으로 사용하려면 상업 라이선스를 구매해야 했습니다.
하지만 이제는 필수가 아니라 선택 사항입니다. 다만, 조직이 Obsidian을 지원하고 싶다면 여전히 라이선스를 구매할 수 있습니다.
이는 개인 사용자용 Catalyst 라이선스와 유사한 방식입니다.

또한, Obsidian을 지원하는 조직은 ‘Obsidian Enterprise’ 페이지에서 소개됩니다.
귀사의 로고를 여기에 표시하려면 25개 이상의 라이선스를 구매하면 됩니다.

Obsidian은 100% 사용자 지원으로 운영됩니다.

Obsidian은 여전히 Sync, Publish와 같은 추가 서비스를 통해 운영됩니다.
미래에는 팀을 위한 추가 서비스를 제공할 계획이지만, 이 역시 선택 사항으로 남을 것입니다.

💜 Obsidian을 사랑해 주셔서 감사합니다!