Following up on the Snowflake/MCP discussion that I posed in Integrations, I put together an open-source starting point for teams that want to connect approved Gong data to AI assistants like ChatGPT, Claude, Cursor, or other MCP-compatible tools.
gongctl is an unofficial, local-first Gong API CLI plus a read-only MCP server. It syncs approved Gong data into your own SQLite or Postgres cache, then exposes a controlled MCP surface that an assistant can query without receiving Gong API credentials, raw exports, writable database access, or full transcript dumps.
The goal is not to replace Gong AI, Smart Trackers, scorecards, or Gong-native workflows. I see it as a launch point for focused point solutions that can enhance current Gong offerings: objection discovery, tracker or scorecard candidate analysis, coverage checks, campaign or enablement research, and other evidence-backed workflows that can eventually be operationalized back in Gong.
I am looking for feedback from teams using Gong heavily, especially RevOps, marketing, sales enablement, and analyst teams. I will also help the first 3 companies that want to try it set up a controlled pilot.
GitHub: https://github.com/fyne-coder/gongcli_mcp
What would you want ChatGPT or Claude to answer from your Gong call evidence if the access path was read-only and customer-hosted?