Context7 with llama.cpp WebUI

Install + handshake last tested 2026-10-22 · Local-model tests: Not yet tested

Client: llama.cpp llama-server WebUI (--webui-mcp-proxy). Client versions and features come from vendor docs, checked 2026-10-22. We haven't run this setup end-to-end yet.

Prerequisites

• A recent llama.cpp build that includes --webui-mcp-proxy
• A GGUF model with tool-calling support
• Node.js 18+ if you run Context7 locally

Steps

1. Start llama-server with your model and --webui-mcp-proxy.
2. In the WebUI MCP settings, add the Context7 server.
3. Context7 has only 2 tools, so no filtering is needed.

Config

{
  "mcpServers": {
    "context7-upstash": {
      "command": "npx",
      "args": [
        "-y",
        "@upstash/context7-mcp"
      ]
    }
  }
}

Suggested tool allowlist

Picked from the real tool names in our test; whether a small model succeeds with them is not yet tested.

• Context7 (Upstash) (2 tools listed): resolve-library-id, query-docs

Measured context cost

• Context7 (Upstash): ~1,074 tokens (13% of 8k, 3% of 32k) for all 2 tools. Small-model friendly: medium.

Method: tiktoken cl100k_base over compact JSON of name+description+inputSchema (approximate). Cost for the allowlist subset alone: not yet measured.

Success rates on small models

• Context7 (Upstash): Not yet tested

These were our own task runs, with all tools exposed. The setup on this page (client, model, allowlist) hasn't been run end-to-end.

Known failure modes (community-reported, not measured by us)

• Too many tools: the model picks the wrong one. Fix: use the allowlist above.
• Gemma thinking mode suppresses tool calls. Fix: turn thinking off.
• Context overflow from large tool schemas. Fix: fewer tools, or a bigger context window.

FAQ

Does Context7 (Upstash) work with local models?

Not yet tested. We have only run our standard install and handshake tests (last tested 2026-10-22); tool-call success with local models has not been measured.

How many tools does it expose?

Context7 (Upstash): 2 tools in our tools/list test.

Why limit the tools?

Community reports say small models choose the wrong tool or emit malformed arguments when shown many tools. We haven't measured this yet.

Are there affiliate links on this page?

No.