Point your agent at the boardroom.
BoardHerald speaks the Model Context Protocol. Any MCP-compliant agent — Claude Desktop, ChatGPT, or your own — can read and write board material on your behalf, scoped to the same permissions you'd have yourself.
The four pages below cover everything from copy-paste setup to deep configuration and troubleshooting. Start at Quick start if you just want a working Claude Desktop integration in the next five minutes.
New to the Model Context Protocol? modelcontextprotocol.io is the spec. MCP is an open standard for letting agents call tools on an external service, maintained by Anthropic and adopted by a growing number of clients.
In this section
Copy-paste configs for Claude Desktop, ChatGPT, and raw MCP clients.
open →Personas, scopes, and what each agent tier can read or write.
open →API keys, the bridge, env vars, and self-hosting notes.
open →Push KPI period values from Stripe, a webhook, an MCP cron, or a CSV upload.
open →Error codes, common gotchas, and where to file an issue.
open →