Well-crafted Model Context Protocol servers — production-grade, typed, and read-only by default.
mcpwright (noun) — a maker of MCP servers. Each repo here brings a real-world data source
into your agent through a tight, well-documented tool surface: clear descriptions the model can
actually reason about, structured (typed) returns, and safe-by-default behavior.
Built on Anthropic's official mcp Python SDK.
| Server | What it does | Status |
|---|---|---|
| edgar-mcp | SEC EDGAR — resolve issuers, search filings, and pull recent Reg CF (Form C) / Reg D (Form D) offerings, right inside your agent. | 🚧 in progress |
More on the way.
- Official SDK. Python,
mcp.server.fastmcp— the canonical Anthropic stack. - Typed, structured output. Tools return Pydantic models, so agents get structured data, not just text.
- Read-only by default. Tools are annotated (
readOnlyHint) and safe to run; clear, actionable errors. - Easy to run.
uvx <server>-mcp, with copy-paste config for Claude Desktop and Claude Code. - Discoverable. Published to PyPI and the official MCP Registry.
Built by Devender Gollapally — Principal Software Engineer shipping production agentic AI in regulated finance, and ERC-1450 co-author. devender.me · LinkedIn
Contributions and issues welcome on any repo. Always happy to talk applied AI, MCP, and on-chain securities.
