Skip to content
All articles
Engineering6 August 2025·4 min read

Understanding MCP features: Tools, Resources, Prompts, Sampling, Roots, and Elicitation

The uncomfortable part of building for agents is that the old assumptions quietly stop holding.

Paycux engineering

The uncomfortable part of building for agents is that the old assumptions quietly stop holding. There is no browser to redirect, no person to read a consent screen at the moment it matters, and no obvious place to put the word "no".

This piece walks through how we think about it at Paycux, what we have changed our minds about, and where the sharp edges are.

Part 1: Building MCP servers

Part 1: Building MCP servers deserves its own treatment. The audit trail matters more here than in any human flow. When something goes wrong, the question is never "did the user intend this" in the abstract — it is which agent, acting under whose authority, made which call, and whether the record can prove it.

Consent is the part most implementations get wrong. Asking once at install time and then acting indefinitely is not consent; it is a standing grant with no expiry and no visibility.

Tools: What your server can do

That brings us to tools: what your server can do. The audit trail matters more here than in any human flow. When something goes wrong, the question is never "did the user intend this" in the abstract — it is which agent, acting under whose authority, made which call, and whether the record can prove it.

The audit trail matters more here than in any human flow. When something goes wrong, the question is never "did the user intend this" in the abstract — it is which agent, acting under whose authority, made which call, and whether the record can prove it.

  • Scope every agent credential to one principal and one task
  • Give tokens minutes of life, not days
  • Record which agent acted, under whose authority, on what
  • Make revocation a single call that takes effect immediately

Resources: What your server knows and shares

That brings us to resources: what your server knows and shares. An agent's credential should describe what it may do, not who owns it. Scope it to the narrowest set of operations that make the task possible, bind it to a single principal, and give it a lifetime measured in minutes rather than days.

Consent is the part most implementations get wrong. Asking once at install time and then acting indefinitely is not consent; it is a standing grant with no expiry and no visibility.

Prompts: Reusable templates for your server

That brings us to prompts: reusable templates for your server. The audit trail matters more here than in any human flow. When something goes wrong, the question is never "did the user intend this" in the abstract — it is which agent, acting under whose authority, made which call, and whether the record can prove it.

An agent's credential should describe what it may do, not who owns it. Scope it to the narrowest set of operations that make the task possible, bind it to a single principal, and give it a lifetime measured in minutes rather than days.

Consent is the part most implementations get wrong.

Bringing it all together: Multi-server coordination

Bringing it all together: Multi-server coordination is where this gets concrete. Consent is the part most implementations get wrong. Asking once at install time and then acting indefinitely is not consent; it is a standing grant with no expiry and no visibility.

An agent's credential should describe what it may do, not who owns it. Scope it to the narrowest set of operations that make the task possible, bind it to a single principal, and give it a lifetime measured in minutes rather than days.

Part 2: Building MCP clients

Consider part 2: building mcp clients. An agent's credential should describe what it may do, not who owns it. Scope it to the narrowest set of operations that make the task possible, bind it to a single principal, and give it a lifetime measured in minutes rather than days.

The audit trail matters more here than in any human flow. When something goes wrong, the question is never "did the user intend this" in the abstract — it is which agent, acting under whose authority, made which call, and whether the record can prove it.

Sampling: Letting MCP servers use your AI model safely

Consider sampling: letting mcp servers use your ai model safely. An agent's credential should describe what it may do, not who owns it. Scope it to the narrowest set of operations that make the task possible, bind it to a single principal, and give it a lifetime measured in minutes rather than days.

The audit trail matters more here than in any human flow. When something goes wrong, the question is never "did the user intend this" in the abstract — it is which agent, acting under whose authority, made which call, and whether the record can prove it.

Where this leaves us

The pattern repeats across every system we have looked at: the hard part is not the mechanism, it is keeping the mechanism honest as the surrounding assumptions change.

If you are working through the same problem and want to compare notes, the docs cover the mechanics and the console shows the behaviour on your own data.

Everything here, already built

Sign-in, enterprise SSO, directory provisioning, roles and an audit trail behind one API. Start with the quickstart and have a working sign-in this afternoon.

Start selling to enterprise customers

Create an account, point sign-in at Paycux, and get back to the part of the product that is actually yours.