Comparisons · How agents get new powers
A plugin upgrades one app. MCP upgrades every agent at once.
For twenty years, giving software a new trick meant writing a plugin (a VS Code extension, a browser add-on) bolted into one host and useless in any other. AI agents inherited that model, then outgrew it in a year. The Model Context Protocol (MCP) is the alternative: a neutral standard that lets one server hand the same capability to every agent at once. Here is the difference, drawn out, and why it is the reason a single install can reach Claude Code, Cursor, and seventeen more.
The two models
Bolted into one app. Or spoken by all of them.
A plugin lives inside its host: it uses that app's private extension API, ships in that app's store, and disappears the moment you open a different editor. Want the same capability in another agent? Build it again. MCP flips the ownership: the capability lives in its own small server, outside any app, and every agent that speaks the protocol connects to the same one. Build once; reach all of them.
The left side is why tool makers used to pick a favorite app and ignore the rest: every new host meant a fresh integration to write and maintain. The right side is why that math changed: the server is written against the protocol, not against any one app, so the app on the other end is interchangeable.
Side by side
Six questions, two very different answers.
| The question | Plugin / extension | MCP server |
|---|---|---|
| Who can use it | One host: the app it was written for | Every MCP agent on the machine |
| How many times it's built | Once per app: a rebuild each time | Once |
| A new agent ships | Write a whole new plugin | It already works |
| You switch editors | Left behind | Comes with you |
| Who owns the surface | The host vendor's private API | An open, shared standard |
| Where it runs | Inside the app | Its own process, on your machine |
None of this makes plugins bad: a deep, app-specific extension is still the right tool when you want to reshape one editor. But when the job is handing a capability to whatever agent you happen to be using, the plugin model asks you to place a bet on which app wins. MCP lets you skip the bet.
The two anchors
Claude Code and Cursor speak the same protocol.
The two agents most people reach for prove the point. A Claude Code setup cannot run inside Cursor; a Cursor extension cannot run inside Claude Code: different apps, different plugin systems, no shared ground. But both speak MCP natively, so one MCP server answers an identical call from either, and from every other MCP client on the machine. Neutrality is not a feature you switch on; it is the default.
The model in each app decides when to call the tool and picks the right one from a plain-language menu. The server does the work and hands back a result. Swap the agent on top and nothing underneath changes: same server, same four tools, same answer.
The file-search MCP server
Interlinked Files is one server every agent shares.
It exposes four tools (filename search, content search, both at once, and a knowledge lookup that returns everything known about one file, folder, or repo) over the same protocol Claude Code and Cursor already speak. There is no VS Code extension to sideload, no Cursor plugin, no per-app anything: one signed installer configures 19 AI clients in a single step, and the moment a new MCP agent ships, it works too. Backed by an always-fresh index of every file on your PC (every repo, every document, even the git-ignored files a project-scoped plugin never sees), measured on a real 4.47-million-file machine. Save a file and it is searchable again in about a millisecond, so the answer is never stale.
ONE SERVER TO BUILD. NINETEEN AGENTS THAT GET IT. ZERO PER-APP PLUGINS.
Measured, not claimed
What one shared server buys.
| The searcher | What ran | Time | vs Interlinked |
|---|---|---|---|
| Interlinked Files | one indexed filename lookup | 85 µs | baseline |
| VS Code (Ctrl+P) | same file, same drive | 35 s | 412,000× |
| Windows Search | median of 20 queries | 67 s | ~480,000× |
| ripgrep | same content query, same corpus | 93.8 s | 58,625× |
| Claude Code (no index) | one file, agentic archaeology | 3 m 13 s | 2,276,000× |
Methodology: Ryzen 9 9950X3D · 64 GB DDR5 · NVMe · Windows 11 · 4,470,000 files. Filename lookups are microsecond-scale (85 µs single, 139 µs median of 20 queries). The ripgrep row is the same content query, content-to-content (93.8 s → 16 ms; 58,625× across the five-query set). The Claude Code row is a measured single-file agent run: the agent, given no index, listing directories until it found the file. Every one of those numbers is identical no matter which of the 19 clients made the call.
Do this today
Skip the plugins. Install the layer.
Install once. One signed installer auto-configures 19 AI clients: Claude Code, Cursor, Copilot, Codex, Windsurf, Zed, Cline and the rest. No extension to sideload, no JSON to edit.
Let it index. Filename search works immediately; content search fills in behind it. At rest the whole engine sits around 44 MB, less than a browser tab.
Ask from any agent. Whichever agent you happen to be in, the tools are already there and answer identically. Switch editors tomorrow and there is nothing to reinstall.
It stays fresh. Save a file and it's searchable in about a millisecond, under 30 in the worst case. The index rides your changes; you never re-index by hand.
Everything local is free forever, on all your devices, no card. Hosting starts at $5.99/mo. Download for Windows.