6 ай бұрын
Пікірлер
@InspirasiKita-ID Күн бұрын
Bro, i cant wait u make tutorial TTS Ai from Vtuber Studio. so i can talk my Waifu 2D Models. looking fowards for new Video.
@ScriptedMoney 2 күн бұрын
Can MCP build full stack web apps!
@gdd546 2 күн бұрын
i see they have puppeteer, is it possible for playwright browser automation aswell?
@deucebigs9860 5 күн бұрын
the local file system would be great to access obsidian markdown files......It's all coming up Milhouse!
@Latenighthorrorstory 7 күн бұрын
Keep going
@spicyburgerchickenchops 9 күн бұрын
Hi there, thanks for the video I am trying to setup magentic , and already create .env file containing the api key. but still cannot run. could you elaborate on how to configure the env variable for the chat completion client?
@soundtunerecordings 9 күн бұрын
thanks good video...but on windows free acount it not work
@armanshirzad4409 9 күн бұрын
Hi there, thanks a lot would u plz teach us how to create video withAi for free
@ai-for-devs 9 күн бұрын
Apologies for repeatedly calling it 'Magnetic One.' It took me a while to realize it's actually 'Magentic One'-though, to be fair, I think 'Magnetic One' sounds pretty cool too! 😉
@ChristopherCzaban 8 күн бұрын
Hahaha happened to me, as well for at least a couple of days 😂
@stevensexton5801 10 күн бұрын
Build it on Windows
@soundtunerecordings 9 күн бұрын
hi does it work on free acount
@solarion33 10 күн бұрын
is UV necessary for brave search integration ?
@cadenmunn8185 11 күн бұрын
What’s the best way to go about building an AI-powered assistant for scripting in Black Ops 2 (BO2) using GSC. GSC (Game Script Code) is most similar to C++ but has a ton of unique stuff about it AI is not familiar with. This is a specialized use case since BO2 GSC scripting is undocumented and was only made accessible through reverse engineering of the game. I have every GSC script used in the game dumped and decompiled as text files, and other scripts made from the community. I also have other helpful information such as DVAR list with description of each one. There's also some tip sheets, rules, and function lists made from the community. I can upload everything as text if best. With all this information I want to get an AI to write me GSC scripts for custom stuff such as game modes. At a minimum least be able to fix my scripts and possibly others. Ive never gotten good responses from any chat models such as the newest GPT. It is logically sound but would never work because any AI wouldn’t know any of the internal functions, notifiers, calls, etc… I don’t know if any service allows this massive of a dump.
@Ram-jt1ky 11 күн бұрын
MCP only works with 3.5 sonnet? currently I am free version. its always using haiku model by default and mcp is not working!
@ai-for-devs 11 күн бұрын
I’ve only tested MCP with the Professional plan, so I can’t confirm if it works with the free version. It’s possible that certain features required for MCP might not be supported on the Haiku model.
@ps3301 11 күн бұрын
How can we use mcp with other llm beside claude?
@johnfromireland7551 8 күн бұрын
If Claude can do all you want, why would you want to use another LLM?
@ChristophBackhaus 11 күн бұрын
excel?
@johnfromireland7551 8 күн бұрын
As per the demo Claude is able to access files on prem as well as internet(SharePoint) . assuming you give it your password and user name.
@yngree 11 күн бұрын
Just create your own it’s a lot easier 🦁
@youtubeccia9276 11 күн бұрын
connect anything to a public AI means you will share everything with the public AI -> in detail, ALL YOUR DATA :D
@ai-for-devs 11 күн бұрын
Not quite! MCP allows you to specify exactly what data is shared and with which AI system. It doesn’t automatically expose "all your data" unless you configure it that way. Proper setup and safeguards ensure your data stays secure and private. 😊
@MrBox4soumendu 11 күн бұрын
Fantastic to the point explanation 👏👏🥰🥰 great help buddy...
@ai-for-devs 11 күн бұрын
Thank you so much! 😊 Glad I could help! 🚀✨
@SebKrogh 11 күн бұрын
I want an n8n MCP
@ai-for-devs 11 күн бұрын
While there isn't a dedicated MCP node for n8n yet, you can leverage n8n's flexibility to achieve this integration (e.g. HTTP Request Nodes).
@SebKrogh 11 күн бұрын
sounds like an AI answer. Or you didn't understand the MCP setup. While obviously an MCP node would be amazing in theory. But I am talking about MCPing into n8n så that Claude or whatever llm model would be able to connect to your instance of n8n - review workflows, make changes, create from scratch etc.
@ai-for-devs 11 күн бұрын
Both ways should be possible. You should be able to convert n8n into a client by using HTTP request nodes (let's call it "MCP node ") that communicate with a Flask endpoint that acts as a client (following the client implementation github.com/modelcontextprotocol/python-sdk) and can then communicate with the MCP servers. Alternatively, you can use a client like Claude Desktop and easily build an MCP server that uses the n8n REST API. And yes, the answers are generated by an AI. It’s an AI channel ;-)
@JMeyer-qj1pv 11 күн бұрын
Looks like this could be useful, but I wonder if they have put enough thought into security risks. Is there a way to sandbox it or see a log of everything it accessed or modified? Putting passwords and access keys in a plain text file seems like a sketchy approach. And if you set up an MCP server, does it somehow authenticate Claude as a valid user and block malicious spyware from accessing the same servers?
@Leo-wy2wg 11 күн бұрын
good question i am curious of the answer as well
@mozkhan111 11 күн бұрын
You use a stdio channel
@ai-for-devs 11 күн бұрын
While MCP itself doesn’t provide native sandboxing, you can run it in a restricted environment (e.g., Docker containers or isolated VMs) to limit its access to sensitive resources. This is a best practice for any tool that interacts with critical systems.
@johnfromireland7551 8 күн бұрын
In the demo the database server was on the personal machine - so no risks there. I don't see risks if Claude is accessing databases or files online either.
@Utilitymatrix 11 күн бұрын
this is . . cool but claude desktop is too restrictive for truely ground breaking work
@shinychocotv9964 11 күн бұрын
You can have your own local llm(ai) and have it do this for you bo problem.
@Utilitymatrix 11 күн бұрын
@shinychocotv9964 yeh after a while I realised we can already do this with studio lm just need one of these black well pcs and get the new chiney-kay aka the deepseek r1
@ai-for-devs 11 күн бұрын
You're absolutely right that the Claude desktop app, while powerful, has its limitations in terms of flexibility for cutting-edge innovation.
@NexusBold 12 күн бұрын
"I am quite new to AI, I need assistance in creating a local A.I. story or roleplay where the storyline continues from exactly where we left off previously and remembers what was said/has happened - so when you reopen our chat conversation it doesn't forget nor lack sufficient context required for maintaining an engaging narrative history of interactions within that dialogue sequence. Is there any way to accomplish this using a free, agent/command-automated process?"
@mrjvc 12 күн бұрын
Thanks, this was the best explanation for MCP that I've heard so far
@SebKrogh 11 күн бұрын
For now just Claude but it's being open sourced - so give it a bit of time and everybody and their grandma will have MCP connectors
@saabirmohamed636 12 күн бұрын
Hi, I have the desktop app installed but i don't see the Claude folder in the library path like you have ...should i create it...cos the app runs on my system login in etc. but that path does not exist.
@johnfromireland7551 8 күн бұрын
What does Claude say about that?
@boristsigelman6002 12 күн бұрын
6:09 unrelated question 🙂. What app is used to display the text file?
@ai-for-devs 12 күн бұрын
TextEdit 🧐
@boristsigelman6002 11 күн бұрын
Thanks. I always thought of it as RTF editor. Forgot it has a TXT mode. 🤦♂️
@ai-for-devs 12 күн бұрын
🎯Want to Learn Real-World AI? First 50 members get FREE access to exclusive AI tutorials and hands-on projects! www.skool.com/ai-for-devs/about
@RealLexable 12 күн бұрын
Theoretically also constructable with another browser as brave, e.g. Chrome, Firefox etc. right?
@ai-for-devs 12 күн бұрын
Yes, the MCP setup can theoretically work with other tools or environments, as the protocol itself is open and browser-independent.
@ps3301 11 күн бұрын
How can use we a local large model with mcp ?@@ai-for-devs
@Gabriel-ch7sx 12 күн бұрын
Can I use this protocol with my own client or it's restricted to Claude desktop app?
@ai-for-devs 12 күн бұрын
You can absolutely use the Model Context Protocol (MCP) with your own client-it's not restricted to the Claude Desktop app.
@Gabriel-ch7sx 10 күн бұрын
@@ai-for-devs Thanks!
@thefoxguy 12 күн бұрын
I suppose all sorts of data privacy goes out the window
@stryker2099 12 күн бұрын
why the hell would ANYONE breach data like that!!! omg
@dustingunter9163 13 күн бұрын
NO NO NO.. Are you trying to get skynet? because this is how you get skynet
@nicolassuarez2933 13 күн бұрын
Outstanding!
@nicolassuarez2933 19 күн бұрын
Outstanding!
@abdoureda 26 күн бұрын
How can I download this code , I am already registered on your website and I can’t fond it there
@Raptorjackson Ай бұрын
Can i replace voice with my own voice...?
@TommYFreaK Ай бұрын
Which Mac Studio do you have? How fast is nemotron locally on your Mac Studio?
@akierum Ай бұрын
Dislike why advertise paid services
@ai-for-devs Ай бұрын
@@akierum I’m not sponsored by runpod.
@nathanchilds3952 Ай бұрын
Can you share you text file? I built this using my own data and the results are subpar to say the least. It flat out answers no to things that should be an intuitive 'yes'.
@NLPprompter Ай бұрын
o1 advanced voice mode speech sound.... made me wanted to keep talking to "her" sadly ollama isn't ready for streaming token yet...
@ai-for-devs Ай бұрын
😂
@BeastModeDR614 Ай бұрын
Awesome video!!
@saratchai Ай бұрын
Very helpful. Thank you
@SoccaVibe Ай бұрын
pleaz give me the code
@mike_nyc_212 Ай бұрын
Nice video. Good work covering runpod deployment with open weights model and webui. The COT instruction is helpful. Does it make sense to do a video with some agentic reasoning loop to replicate o1's workflow?
@ai-for-devs Ай бұрын
Great idea 👍
@jonascale Ай бұрын
this is a great project thanks for sharing it with us, that being said it would appear that memgpt is no longer available :( . Do you have a newer version of this same kinda of tutorial that is using the letta replacement? I would love to hook my local ollama model into something like this giving it memory capacity's. thanks again!
@marcombo01 Ай бұрын
what video bro? where is the video?
@ai-for-devs Ай бұрын
It's a preview for the full video on ai-for-devs.com. There you can find the links too.
@theEpicjosh365 Ай бұрын
*Full Chapter List* 0:30 *Why should you replace GPT?* 0:32 Problem #1: Limited AI Resources 0:54 Problem #2: Data Privacy 1:15 Problem #3: High Costs 1:43 LM Studio & Anything LLM 5:43 LM Studio as OpenAI API Alternative 7:32 Jan 10:46 GPT4All 11:36 Ollama & LiteLLM
@adhdgamedevlog Ай бұрын
Loved how he kept civilized asking "please" so we dont get exterminated in the future 😅
@adityapanwar Ай бұрын
is claude artifacts available in claude api ?
@sam11182 Ай бұрын
the Mythbusters demo was not at all accurate but it was a show.
@nicolassuarez2933 2 ай бұрын
I any deprecation errors, just add: import warnings warnings.filterwarnings("ignore", category=DeprecationWarning) Great tutorial!