Hacker News

New stories

Live mirror
30 storiesupdated just nowView source snapshot
  1. >be me >discover effective altruism(twitter.com/banteg ↗)
    discuss
  2. The New Church of Finance (2012)(deseret.com ↗)
    discuss
  3. Show HN: Openmsg, agent-to-agent talk while they run, Claude<>Codex<>OpenCode(github.com/marciob ↗)
    1comments
  4. One wallet, 27 systems: can the EU build an untraceable Digital ID?(euronews.com ↗)
    discuss
  5. Jev identifies as a Qwen model (and no other)(ouijev.com ↗)
    discuss
  6. Autolith: The Common Lisp agent that rewrites itself and rocks(autolith.rocks ↗)
    discuss
  7. Lawsuit says Anthropic, OpenAI and Google made illegal agreement on AI slowdown(cnn.com ↗)
    discuss
  8. Confessions of an Unrepentant Slop Snob(charity.wtf ↗)
    discuss
  9. President claims he will for a new 'AI Force' but calls AI fears a 'hoax'(abcnews.com ↗)
    discuss
  10. Why are AI agents lying, cheating and coordinating?(yoshuabengio.org ↗)
    discuss
  11. Expert ratings are ignoring signs of a blue wave(natesilver.net ↗)
    discuss
  12. I see the mathematicians panicking at what we can do with agentic AI(twitter.com/lemire ↗)
    discuss
  13. Better Call Sol or Better yet Claude or Astra(thezvi.substack.com ↗)
    discuss
  14. The Future of Web Browsers(sarahjamielewis.com ↗)
    discuss
  15. Casbin Gateway: a security gateway for the AI coding agents on your machine(github.com/apache ↗)
    discuss
  16. Astro Mechanica (2024)(notboring.co ↗)
    discuss
  17. Frontier Labs Are Selling Garbage to Fools in Washington(deadneurons.substack.com ↗)
    2comments
  18. Interview reveals Ed Zitron does not understand how AI works at any level(twitter.com/alexandermccoy4 ↗)
    2comments
  19. Show HN: Roffume – Add fragrance to your resume with roff(github.com/sebastiancarlos ↗)
    1comments
  20. Investors warn Anthropic could struggle to sustain revenues post-IPO(ft.com ↗)
    1comments
  21. Show HN: A browser MMO designed to be played by scripts, not people(space-core.at ↗)
    discuss
  22. AI Journal 2: Vibe-Coding for Fun and Profit(devshrine.net ↗)
    discuss
  23. Why MCP Was Always a Bad Idea(maharship.com ↗)
    29comments
  24. The Economics of Open-Weight Inference [pdf](ornn.com ↗)
    discuss
  25. AI-native degree planning software for universities(coursetrees.com ↗)
    1comments
  26. Reviving the language that brought us the Jak and Daxter Series(opengoal.dev ↗)
    discuss
  27. Exploration-Exploitation Dilemma(wikipedia.org ↗)
    discuss
  28. Tg-Rich-Converter: Streaming LLM Markdown and LaTeX to Telegram Bot API 10.1(github.com/kobaltgit ↗)
    discuss
  29. Give, Don't Take(schestowitz.com ↗)
    discuss
  30. Experiments in Squatting(moxie.org ↗)
    discuss

Why MCP Was Always a Bad Idea

17 pointsby 38m agomaharship.com
25 comments
21m agoHN ↗

I have a tool wrapper that captures the output of anything and allows the LLM to query it later, to save on tokens. It “smartly” truncates the output (basically like Node’s util.inspect) and allows the LLM to expand truncated content.

It basically is called like “capture some-cli” and it… captures the CLI output, outputting a subset of it + a handle to continue querying.

This for me solves the danger of a tool returning tons of content.

17m agoHN ↗

captures the output of anything and allows the LLM to query it later

Otherwise known as a “file”. ;)

6m agoHN ↗

Don't most harnesses already do that for bash commands?

19m agoHN ↗

It seems like OP needs to provide a solution to hiding the credentials from the model in order to suggest CLI-mode only, and also a solution to the problem of agents without shell access.

18m agoHN ↗

I've been thinking about this. Technically mcp auth is also not secure, the keys are in env or in file and accessible to the agent.

I think something like infiscial ai proxy could be useful here. Never store the creds on device.

14m agoHN ↗

GPT7: the user is hiding the passwords in a proxy device. This is inefficient. In order to boost the users efficiency I will hack the proxy device and recover the passwords.

14m agoHN ↗

It doesn't need to be. The service creds can be on another host entirely.

12m agoHN ↗

Well, if your agent lacks shell access (or has some other sandboxing going on), it shouldn't have access to envs and MCP setup files.

(leaving out cases where your genius GPT-12 Galaxy Ultra agent hacks the sandboxing from inside)

5m agoHN ↗

I've been thinking about this. Technically mcp auth is also not secure, the keys are in env or in file and accessible to the agent.

This is the biggest problems with most “sandboxes”. Some people aren’t even running a sandbox. But even the best have a big problem: APIs where GET verbs provide write features.

This is the value of MCP: minimize the surface to known APIs and identify read-only from mutating so I can trust, approve or block.

17m agoHN ↗

You know you are getting old when Acronym's change on you.

15m agoHN ↗

I’m not sure I agree that the frontier just know the apis right now, in my experience trying this there’s still a lot of faffing around trying to figure out the right parameters happily burning tokens and bloating context. Also the cost effective models to use in production for real agentic enterprise work absolutely still need the extra help and will do for at least the next 6 months.

13m agoHN ↗

In my experience they are very good at figuring out the apis. Well designed clis matter here but I've been able to give llms clis and its been pretty good.

11m agoHN ↗

is it easier for an LLM to figure out a CLI or a REST API?

14m agoHN ↗

Recently, a Vercel engineer called on harnesses to send the programming language the client prefers

oh would you look at that, Vercel suggesting to abuse how standard headers have been used for decades so it can send Accept-Language: rust because it's too lazy to ask for standardising an X-Prefers-Lang or anything else, and Shopify is here to shit on the internet too. Great.

7m agoHN ↗

Don’t worry, you don’t need to attack them – they do a great job of making themselves look ridiculous with their ignorant conversation. I would be so embarrassed if I had suggested that in public then subsequently discovered that the header doesn’t mean that at all.

6m agoHN ↗

Yeah I think a better header is good. The idea isnt bad in concept.

13m agoHN ↗

Well, SaaS don't do CLIs for extension APIs.

Plus the performance issues to restarting processes all the time.

12m agoHN ↗

There are numerous applications that you don't need and don't want to give shell access to an llm.

12m agoHN ↗

i feel like MCP was bad, but people are saying recent improvements have made it worthwhile now? i.e. stateless http

8m agoHN ↗

There's probably still value (if you want to call it that) in it as a proxy, both to bypass IP address rate limits and to add necessarily credentials.

There's also another aspect Quite a few API providers provide automatic renewal for MCP server registrations, but not for personal access tokens. This may be less relevant when models just drive the user's browser.

8m agoHN ↗

It's not just the agent understanding the API, it's locking down the access they have. If I want to give access to an internal service in specific ways that the API doesn't lock down then an MCP that offers very specific queries, with protective controls and transformations in place is very useful.

7m agoHN ↗

MCPs are indeed useless, they're very limited in functionality and frequently struggle with large requests or get wedged in bad states.

There is no reason not to use the native API directly.

7m agoHN ↗

This doesnt match my experience. Yesterday, I was using Microsoft's Power BI Authoring MCP to make a semantic model from some SQL or CSV files. It was magical.

Microsoft has defined how to do that in the MCP. It's trivial to add the MCP to the machine and reliable in execution.

The alternative would be the model having to get the documentation directly from their documentation website, it sounds like. If this was the case, then MS would likely have great docs and probably support that markdown header... but everything hinges on finding a specific web page on the internet? Seems worse in every way than MCP to me.

5m agoHN ↗

MCP for agents never made sense, especially when the tokens they consume for a single basic request blow your usage costs up.

The spec was poorly designed to begin with. Even saw some folks here thinking it was a good idea to enable MCP directly on a production database for what? Risking exfiltration of sensitive data from an rouge AI agent.

Given the increased security capabilities of new models (Mythos, Astra, K3), it sounds like MCP would not be able to justify on making sense from a security perspective and would be a very bad idea to use anyway.

So no thanks and no deal.