procontext 0.1.2

MCP server for accurate, up-to-date library documentation

ProContext

Documentation layer for AI coding agents

Your AI coding agent generates broken code because it doesn't have current docs.

ProContext fixes that - verified library documentation, served on demand.

2000+ libraries · Works with Claude Code, Cursor, Codex, VS Code (GitHub Copilot), Windsurf, and any MCP-compatible tool

---

Why ProContext

	Without ProContext	With ProContext
Discovery	Agent guesses APIs from stale training data	Agent reads from a curated registry of 2000+ libraries
Sources	Web scraping and outdated search results	Hand-picked, verified documentation sources - no scraping, no noise
Freshness	Docs go stale as libraries update	Registry refreshes automatically in the background
Speed	Repeated fetches hit the network every time	Every page cached after the first fetch - millisecond lookups
Context	Entire docs dumped into context	Large pages served in paginated windows to respect token limits
Precision	Agent scrolls through irrelevant content	Search within a page or browse its outline to jump to the right section
Safety	Arbitrary URL access risks data leaks	Fetches restricted to known documentation domains

Quick Start

Install ProContext:

macOS and Linux

curl -fsSL https://raw.githubusercontent.com/procontexthq/procontext/main/install.sh | bash

Windows

Run this in a PowerShell terminal:

irm https://raw.githubusercontent.com/procontexthq/procontext/main/install.ps1 | iex

Add ProContext to your MCP client config:

{
  "mcpServers": {
    "procontext": {
      "command": "uv",
      "args": ["run", "--project", "/path/to/procontext-source", "procontext"]
    }
  }
}

The installer manages a local checkout and prints the path to use with uv run --project .... For manual install, version pinning, and troubleshooting, see docs/cli/installation.md.

ProContext works with any MCP-compatible tool. The Integration Guide has copy-paste configurations for Claude Code, Claude Desktop, Cursor, Windsurf, VS Code (GitHub Copilot), OpenAI Codex CLI, and Amazon Q CLI. Both stdio (local) and HTTP (shared/remote) transports are supported.

Documentation

• Integration guide
• Installation guide
• CLI docs
• API reference
• Technical spec
• Project site

Registry

The library registry is maintained in a separate repository: procontexthq/procontexthq.github.io

If you want to add a library or update an existing entry, open a PR there rather than in this repository.

Contributing

See CONTRIBUTING.md for contributor setup, development workflow, coding conventions, and pull request guidance.

License

MIT - see LICENSE for details.

---

Built with ❤️ for AI coding agents