Hey r/legaltech.
Wanted to share something we built and ask for honest feedback from the people who'd actually use i
Wee open-sourced a Claude + Clio MCP connector this week. MCP = Model Context Protocol, Anthropic's standard for connecting Claude to data sources. 15 tools across matters, contacts, documents, tasks, calendar, and billing.
Local-only (stdio transport, nothing leaves the lawyer's machine). MIT license, free. The differentiator: it's the only open-source Clio MCP connector with ABA Opinion 512-aligned audit logging built in. Append-only JSONL log of every Claude interaction with Clio data, kept locally.
Three other open-source Clio MCP options exist; none ship with audit logging.
Install:
npx u/oktopeak/clio-mcp
Plus 3 env vars in your Claude Desktop config. Five minutes if Node and Claude Desktop are already on the machine.
Repo: github.com/oktopeak/clio-mcp
What I'd actually like feedback on:
- Is the audit log format useful for an ethics review or state bar audit, or are we missing fields someone with ABA 512 review experience would want?
- We kept write scope tight - only create_task and create_note. Conservative enough, or should we add more?
- Region handling: built US/EU/CA support via env var. Anyone using Clio outside US who can test EU/CA?
- What's the most common Clio integration pain that ISN'T solved by this connector?
Thanks.