IMAP Mailbox

In meinGPT (UI)

For most teams, setup is done directly in meinGPT without editing local config files.

Open admin settings in meinGPT
Go to Data Pools / Data Sources
Click Add Source and choose this source type
Configure credentials and scope in the UI
Save and trigger the first sync

If you do not run your own DataVault runtime, this is usually all you need.

On-Prem Runtime Configuration (Advanced)

data_pools:
  - id: support-mailbox
    type: imap
    imap_server: "imap.example.com"
    imap_user: $IMAP_USER
    imap_password: $IMAP_PASSWORD
    imap_port: 993
    imap_use_ssl: true
    folders: "INBOX,Support"
    exclude_folders: "Spam,Trash"

Configuration Options

Field	Type	Default	Required	Description
`id`	string	-	✅	Unique identifier for the data pool
`type`	string	-	✅	Must be `imap`
`imap_server`	string	-	✅	IMAP server hostname
`imap_user`	string	-	✅	IMAP username
`imap_password`	string	-	✅	IMAP password
`imap_port`	integer	`993`	❌	IMAP port
`imap_use_ssl`	boolean	`true`	❌	Enable SSL/TLS
`imap_timeout`	integer	`60`	❌	Connection timeout in seconds
`folders`	string	null	❌	Comma-separated include folders
`exclude_folders`	string	null	❌	Comma-separated exclude folders
`overwrite_existing`	boolean	`false`	❌	Reprocess already synced emails
`strip_attachments`	boolean	`false`	❌	Keep attachment metadata only
`max_messages`	integer	null	❌	Maximum messages per run
`ssl_key_file`	string	null	❌	Optional client key file
`ssl_cert_file`	string	null	❌	Optional client cert file

Synchronization

Vault reads messages from configured folders and stores them as local sync files.
Next sync runs continue incrementally; overwrite_existing controls reprocessing behavior.
You can cap run size via max_messages for controlled rollouts.

Setup

Create a read-only mailbox/service account
Restrict folder scope (folders and exclude_folders)
Start with a low max_messages limit
Validate sync results, then increase limits

On-prem only: this source page is relevant when you operate your own DataVault runtime and configure data_pools yourself.