Hacker News new | ask | show | jobs
by boleary-gl 419 days ago
You’re not wrong but also this does raise a central question that I think is super un-considered in this whole MCP thing: how are we handling identity in those contexts.

If anything we should be more concerned so it that because of the power that it can hand over to agents.
2 comments
adityavinodh 418 days ago
I think this is the more important question too. I don't think it is right in many cases to use the identity of the user and provide access to these agents. If it a simple one-time task, that might work if you can give restricted temporary access to the agent.

But for any other long-term task that may span hours or days while needing access to various data sources or APIs, we need a system where the agent has its own identity (which may be tied to the user). Just as humans are, agents might not function in the ideal manner at all times. So, we might need a system to monitor 'karma' of these agents. That ways API providers can confidently provide access to both humans and agents, and limit their risk to dangerous agents.

link
adamm255 419 days ago
Totally. Still getting my head around this write up but it goes into a lot of detail. https://aaronparecki.com/2025/04/03/15/oauth-for-model-conte...
link
slowmovintarget 419 days ago
Following those guidelines, how do you not end up with a perpetual 401 response from the REST API?

I understand the idea of separating the OAuth audience between the MCP Server and the REST API it wraps. What I don't understand is how the MCP Server itself gets authorized against the REST API, unless there's a privileged client (that is the MCP Server has an API client by which it identifies itself, and not the end user).

How do you operate within the privileges of the end user in that case? It seems like it would still require the REST API to accept some additional signal of the end user's identity in order to make the authorization decisions. So while the MCP Server access token is "no good on the REST APIs" you have the additional problem of either "trust me, I'm an MCP Server" or the MCP Server has to exchange the "no good" token for an equivalent "good" token that also somehow carries the index to limitations of the user (identity in the case of fine-grained access control, and scopes in the case of coarse-grained).

link