Product Documentation

OpenMind Core documentation

This page is the pre-login guide for setup, security defaults, and day-one rollout so teams can move from raw exports to reliable memory quickly.

Open the app Compare plans

Quickstart path: capture, import preview, dedup check, semantic search.
Auth options: session, API key, OAuth 2.0, and Basic auth.
MCP server endpoint with reusable client snippets.
Checklist-driven onboarding for admins and members.

1) Core architecture

OpenMind Core stores team memory in PostgreSQL with typed entries, metadata, tags, and embeddings.

The API serves both dashboard features and MCP retrieval flows from the same source of truth.

Single workspace memory index
User-scoped dedup guardrails
Semantic retrieval for cross-tool context reuse

2) Import workflow

Upload CSV, TXT, JSON, or JSONL exports. Run preview before commit to validate rows and mapping.

Only import approved rows so low-quality history does not degrade future retrieval quality.

Preview parsed rows and inferred types
Adjust CSV column mapping before import
Confirm duplicate counts before writing

3) Auth and security defaults

Use session auth for dashboard access and API keys or OAuth for external MCP clients.

Set token and session lifetimes in settings to match your security posture.

Session cookie defaults with HTTP-only storage
Revokable API keys per integration
OAuth 2.0 endpoints for browser-based client login

4) Onboarding sequence

Complete the in-app checklist to capture first memory, configure auth, and validate initial operations.

Admin users should also create at least one teammate account to test role-scoped behavior.

Capture first memory
Generate first API key
Register OAuth client (optional but recommended)
Review session lifetime settings

Need hands-on custom channels, persona behavior, or done-for-you workflow automation? Keep Core as the memory layer and add the Agent Add-on when that scope is justified.