Blog
Notes on Toolport, MCP, and local-first AI tooling.
-
MCP on local models: what actually works
Local models pay the highest price for MCP tool bloat and get the least honest advice about it. Here is what works today with LM Studio, Jan, Goose and Open WebUI, what a 7B model can and can't do, and when to use lazy discovery versus the full catalog.
-
Tool bloat doesn't just cost money. It makes your agent dumber.
Every MCP server you connect dumps its full tool list into context on every turn. That isn't only a bigger bill, it's your model spending its sharpest attention on a list of tools before it reads your request. Here is why that degrades the whole system, with the graded proof that fixing it costs you nothing.
-
What it takes to sit in the middle of everything
The earlier posts made the case for lazy discovery. This weekend was the less glamorous follow-on: getting Toolport to reach more clients, hardening the part that holds your credentials, and cutting the setup down to a few clicks.
-
Same answers, up to 91% fewer tokens: we measured it on a frontier model
A graded benchmark of flat MCP tool exposure vs Toolport's lazy discovery on GPT-5.5. Every task completed correctly in both modes; lazy used 74-91% fewer tokens, and the gap grows as you add servers.
-
Context is becoming expensive real estate
Token cost used to be a developer footnote. As agents get deployed for real, it turns into infrastructure cost, latency, and margin. Notes on a category that is starting to form.
-
Your agent doesn't need 500 tools. It needs 3.
How Toolport keeps an AI agent's context flat no matter how many MCP servers you connect.