191–194 of 194 posts
Earlier quoted context omitted.
What does deterministic mean in this case?
Right, I'm using a perhaps loose definition of deterministic here, as in -- "With AI tools, you cannot 100% predict any output based on input, the way you can with most programming."
Earlier quoted context omitted.
But none of the criticisms here is specific to MCP, just to tool calls in general, it wouldn't matter if the agent used a custom tool protocol, plain OpenAPIs, etc. These issues would still exist.
> These issues would still exist. I just explained why these issues don't apply to the LLM writing and invoking code: again, this is because code can apply successive transformations to the input without having to feed the intermediate results into the LLM's context. That code can read a file that would weigh in at 50,000 tokens, chain 100 functions together, producing a line that would be 20 tokens, and only the LLM…
You need a way to expose tools to LLM, even for what you're talking about. The LLM can't write code without access to code writing tools.
MCP is a protocol for exposing tools to an LLM agent in an extensible way.
If you didn't have MCP for this, the alternative wouldn't be a python script. It would be some other tool calling protocol.
Maybe a good criticism is that the protocol doesn't have a defined mechanism for piping tool calls together. It be nice to define as part of the protocol a standard way for the model to say call tool1 and then pass output.foo as input.bar of tool2.
I feel that can probably be an eventual extension.
As an advanced user of coding agents that you let run locally or have sandboxed yourself, you might get more power and mileage by having it implement scripts and execute them, or use bash/curl and so on, to say access the Google Drive standard APIs for whatever you want to do, than to use the Google Drive MCP. I 100% agree. But that's not adequate for all users, and it's not really an alternative to supporting custom tools in an agent.
Earlier quoted context omitted.
Chrome Devtools is similarly an extremely high value MCP for me. I would agree that if you don't find they add discoverability then MCPs would have no value for you and be worse than cli tools. It sounds like we have had very opposite experiences here.
Interesting. Perhaps it comes down to which platforms we're working on. I don't want to be outright dismissive of it. My primary platform is Claude Code. Are you working with another driver e.g. OpenAI Codex?
Actually my primary value is emacs integration with claude.
I have an mcp with one function ( as_user_eval ) which allows it to execute arbitrary s-expressions against the portal package for emacs.
I use this often with custom slash commands, i.e., `/read-emacs`, which instructs claude to use that mcp to pull the context from multiple pseudoregions in to the context window (along with filenames and line numbers). This saves me from having to copy paste all of that.
I understand what the others are saying but the using the portal to talk to a running emacs client I don't find to be particularly "discoverable" on the cli.
I can say things like, "show me in emacs the test that failed", or, "highlight the lines you are talking about", or, "interactively remap my keybindings to do X", or, "take me to the info page that covers this topic".
This, paired with chrome devtools and playwright, had been a real productivity booster for me, and is quite fun.
I use voice dictation for this so it feels like I'm in Star Trek :)
I'm sure in 10 minutes we will be onto the next version of MCP... skills, ACP, Tools, whatever.
But this extensibility/discoverability for me has been nice. I make no stronger claims than that about "what it is for" or "what is should be", as I am a simple hacker with simple needs.
Earlier quoted context omitted.
That is an easily falsifiable statement. If I ask ChatGPT or Claude what MCP is Model Context Protocol comes up, and furthermore it can clearly explain what MCP does. That seems unlikely to be a coincidental hallucination.
Training data =/= web search Both ChatGPT and Claude will perform web searches when you ask them a question, which the fact that you got this confused is ironically topical. But you're still misunderstanding the principle point because at some point these models will undoubtedly have access to that data and be trained on it. But they didn't need to be, because LLM function & tool calling is already trained on these m…
I think you're making a weird semantic argument. How is MCP use not a tool call?