Making MCP cheaper via CLI

#1

Post by thellimist » Wed, Feb 25, 2026, 8:29 PM UTC

Re: Making MCP cheaper via CLI

#2

Post by crooked-v » Wed, Feb 25, 2026, 8:52 PM UTC

Cheaper, but is it more effective?

I know I saw something about the Next.js devs experimenting with just dumping an entire index of doc files into AGENTS.md and it being used significantly more by Claude than any skills/tool call stuff.

Re: Making MCP cheaper via CLI

#3

Post by bdavbdav » Wed, Feb 25, 2026, 8:55 PM UTC

I’m not sure how this works. A lot of that tool description is important to the Agent understanding what it can and can’t do with the specific MCP provider. You’d have to make up for that with a much longer overarching description. Especially for internal only tools that the LLM has no intrinsic context for.

Re: Making MCP cheaper via CLI

#4

Post by speedgoose » Wed, Feb 25, 2026, 9:09 PM UTC

MCP has some schemas though. CLI is a bit of a mess.

But MCP today isn’t ideal. I think we need to have some catalogs where the agents can fetch more information about MCP services instead of filling the context with not relevant noise.

Re: Making MCP cheaper via CLI

#5

Post by mijoharas » Wed, Feb 25, 2026, 9:15 PM UTC

This sounds similar to MCPorter[0], can anyone point out the differences?

[0] https://github.com/steipete/mcporter

Re: Making MCP cheaper via CLI

#6

Post by thellimist » Wed, Feb 25, 2026, 9:16 PM UTC

speedgoose wrote: post #4

MCP has some schemas though. CLI is a bit of a mess. But MCP today isn’t ideal. I think we need to have some catalogs where the agents can fetch more information about MCP services instead of filling the context with not relevant noise.

It's the same from functionality perspective. The schema's are converted to CLI versions of it. It's a UI change more than anything.

Re: Making MCP cheaper via CLI

#7

Post by red_hare » Wed, Feb 25, 2026, 9:16 PM UTC

True for coding agents running SotA models where you're the human-in-the-loop approving, less true for your deployed agents running on cheap models that you don't see what's being executed.

But yeah, a concrete example is playwright-mcp vs playwright-cli: https://testcollab.com/blog/playwright-cli

Re: Making MCP cheaper via CLI

#8

Post by dang » Wed, Feb 25, 2026, 9:20 PM UTC

The article's link to clihub.sh is broken. Looks like https://clihub.org/ is the correct link? I've added that to the toptext as well.

Edit: took out because I think that was something different.

Re: Making MCP cheaper via CLI

#9

Post by thellimist » Wed, Feb 25, 2026, 9:20 PM UTC

bdavbdav wrote: post #3

I’m not sure how this works. A lot of that tool description is important to the Agent understanding what it can and can’t do with the specific MCP provider. You’d have to make up for that with a much longer overarching description. Especially for internal only tools that the LLM has no intrinsic context for.

I can give example.

LLM only know `linear` tool exists.

I ask "get me the comments in the last issue"

Next call LLM does is

`linear --help 2>&1 | grep -i -E "search|list.issue|get.issue")` then `linear list-issues --raw '{"limit": 3}' -o json 2>&1 | head -80)` then `linear list-comments --issue-id "abc1ceae-aaaa-bbbb-9aaa-6bef0325ebd0" 2>&1)`

So even the --help has filtering by default. Current models are pretty good

Re: Making MCP cheaper via CLI

#10

Post by andybak » Wed, Feb 25, 2026, 9:21 PM UTC

Why are they using JSON in the context? I thought we'd figured out that the extra syntax was a waste of tokens?