Local-first files, outbound-only connections, and a human gate on anything that writes.
Where your data lives
AI-specific leak paths
AI agents read real data and can take real actions. These are the honest leak paths, and the control on each one.
Every MCP server runs in its own process over stdio — no shared memory, with timeouts and OS-level limits.
checkpoint nodes pause workflows for a person, and agent permission prompts gate every write action.
Owner, editor, or viewer per doc — read-only links and time-limited access when you share.
Point agents at Ollama or LM Studio and nothing leaves the machine — or bring your own API keys.
Workflow auth pulls tokens from env or the OS keychain — .workflow.md stays safe to commit.
Every run, step, and tool call is logged locally — and every save is a version you can restore.
Found something? Report it via support.