Internet Context MCP Server

Read-only MCP server giving AI agents the web as compact, ranked, verified evidence. Local cross-encoder reranker, NLI claim verification, semantic cross-source agreement and contradiction detection. No API keys. The internet context mcp mcp server wraps that behind the Model Context Protocol, so an assistant can use it through 6 defined tools rather than through you.

What it actually does

Six read-only tools — web_research, web_context, web_search, web_read, web_verify, web_extract — plus MCP resources, prompts, and proper outputSchema / readOnlyHint metadata. Behind each tool: local BM25 ranking, a local cross-encoder reranker, a local NLI classifier, local sentence embeddings, a regex prompt-injection scanner, and a two-tier (in-memory + SQLite) fetch cache.

Its toolset

Everything the assistant can do here goes through one of these:

• web_research — One-shot research tool: search the web, fetch the top N results in parallel, rank chunks within each source (with the local reranker by default)
• web_context — Fetches a URL, cleans it, splits it into chunks, ranks chunks against an agent task with a local BM25-style algorithm, and returns only the best
• web_read — Fetches a URL, removes noisy page chrome, extracts the main content, and returns clean text plus token savings metadata
• web_search — Searches the web and returns compact, source-classified results
• web_verify — Checks whether a claim is supported, refuted, or unclear from one or more source URLs. Fetches each source, ranks chunks against the claim, and looks
• web_extract — Best-effort generic field extraction from clean page text. This is intentionally secondary to web_context; in many agents, the better flow is to call

Adding it to your client

playwright on npm is all you need. Most clients run it directly, so configuration is a few lines and a restart.

Configuration

You will need 5 environment variables: BRAVE_SEARCH_API_KEY, INTERNET_CONTEXT_MCP_RERANK, INTERNET_CONTEXT_MCP_NLI, INTERNET_CONTEXT_MCP_EMBEDDINGS, INTERNET_CONTEXT_MCP_CACHE_DIR. The server will not start without them, which is usually why the tools fail to appear on a first run. Keep credentials in your client's env block or a secrets manager rather than in a file you might commit.

When to reach for it

This sits in the planning and project tracking group, where several servers overlap in what they claim to do but differ sharply once you actually set them up. Internet Context MCP's toolset — web_research, web_context, web_read and 3 more — is a fair guide to whether it matches your workflow. It is maintained by vivekvar-dl; worth a glance at recent repository activity before you build anything load-bearing on it.

We check each listing at SyncDev against the project's documentation before it goes live — if something here drifts out of date, it is a bug worth reporting.

Caveats

• It runs with your machine's permissions. That is convenient and also the reason to think about what you point it at before you approve a tool call.
• Missing credentials fail quietly in some clients — if no tools show up, check the environment block first.
• MCP clients confirm each tool call by default. Leave that on until you have watched what the internet context mcp mcp server does with a few real requests.