Every instance is named — org labels show as Instance / Organization across the app.
| Name | Admin | Groups |
|---|
| Name | Companies | Last sign-in |
|---|
No providers yet. Add one below or set VOITTA_GOOGLE_AUTH_CLIENT_ID in .env.
<data_dir>/admin/indexing_caps.json and take effect
on the next file the worker processes — no restart needed.
Bytes accept human suffixes (e.g. 5MB, 1GB).
| Setting | Value | Default | Range |
|---|
Tip: drag folders straight onto the folder area to upload several at once — each keeps its subfolders.
Created under .
The label shown in the sidebar. Cosmetic only.
Renames the directory under .
Search index, embeddings and files are all preserved — no
re-index needed. Blocked while jobs are running.
Uses the ssh-agent and ~/.ssh/config of the
account running this server — the right choice for
hardware keys (YubiKey / SSHCA / touch2ssh), where the
private key can't be exported. Use an SSH repo URL
(git@github.com:org/repo.git or your org's
org-…@github.com:… form). The server must be
started from an environment where ssh-add -l
lists your key; a YubiKey touch/PIN, if required, is
handled by the agent.
Hold ⌘ / Ctrl to select multiple. Leave empty when "all branches" is on.
Native Workspace files are rendered to indexable markdown:
Docs → one .md per tab (with inline images);
Sheets → one .md per sheet (capped at 100 rows) plus the
full workbook stashed under .voitta_workbooks/;
Slides → one .md per slide with thumbnail and speaker notes;
Drawings → PNG; Forms → text. The connector needs read scopes
for Drive / Docs / Sheets / Slides / Forms — enable each of
those APIs in your GCP project. Existing OAuth connections
will be prompted to re-consent the first time they sync after
upgrading.
Connect your Google account with one click — no Google Cloud setup needed. A browser window opens for Google's consent screen; Voitta gets read-only access to the Drive folders you pick below.
This folder authenticates through , a shared across your organization — there is nothing to configure here. To use different credentials for this folder, switch to another tab and enter them there.
Setup: in Google Cloud Console, create an OAuth
client of type "Web application" and add this exact URL
to its Authorized redirect URIs:
…loading….
Then paste its credentials below, click Save,
then Connect.
Register http://localhost:53682/api/sync/oauth/google/callback
in Google Cloud Console and run the local nginx bridge on
this machine. The Voitta hostname never appears in GCP.
Setup: in Google Cloud Console, create a
Service account with Drive read access, generate a
JSON key, and paste the entire file below.
Share each Drive folder with the service account's
client_email (Viewer is enough) — service
accounts only see what's been explicitly shared with them.
Index a folder straight from the Google Drive app on this Mac — no client ID, no OAuth. Files download on demand as they're indexed; nothing is ever written back to Drive. Native Google Docs are added as links (shared ones are also exported and indexed).
Click ▶ to expand. Tick the folders to index — checking a folder covers all its subfolders. The selected folders appear as a tree under this folder, mirroring their Google Drive structure.
Captures Teams meetings (subject, attendees, transcripts when available). Recordings are recorded as deep links — MP4 files are NOT downloaded. Missing scopes are surfaced below.
Register http://localhost:53682/api/sync/oauth/microsoft/callback
in Azure AD and run the local nginx bridge.
Mirrors selected Jira projects — one markdown file per issue, with status, assignee, labels, dates and other attributes captured as filterable fields. Cloud needs your Atlassian email + an API token; Server / Data Center needs a personal access token.
Only issues updated on/after this date are synced.
ANDed onto the project + date filter.
Mirrors selected Confluence spaces — one markdown file per page, with space, labels, version, ancestors and author/date captured as filterable fields. Cloud needs your Atlassian email + an API token; Server / Data Center needs a personal access token.
Only pages modified on/after this date are synced.
ANDed onto the space + date filter.
Pick one or more folders to sync from the admin's NFS root. Parent checkboxes select everything underneath; partial selection shows as a dash. Folder structure is preserved on disk (two siblings won't collide). Symlinks are skipped.
Click ▶ to expand, click a checkbox to (de)select that subtree.
Personal access tokens for programmatic use (MCP, scripting). Each key is shown once at creation — copy it then; you cannot retrieve it later.
This server exposes an MCP endpoint at .
Create an API key below, then register the server with your Claude client,
replacing <YOUR_API_KEY> with the key.
Add the mcpServers entry to claude_desktop_config.json
(Claude Desktop → Settings → Developer → Edit Config) or to a project's
.mcp.json for Claude Code:
Copy this token now — you won't see it again.
For claude_desktop_config.json (Claude Desktop) or .mcp.json (Claude Code):
| Name | Prefix | Created | Last used |
|---|
No keys yet. Create one above.
A company key grants programmatic access for any member of this scope.
Clients must pair it with a user email — an X-Voitta-User-Email header,
or the email appended to the token as cvk_…:user@example.com.
Shown once at creation.
Same MCP endpoint (), but a company key must be
paired with the member's email via the X-Voitta-User-Email header —
swap in each user's own address. Replace <YOUR_COMPANY_KEY>
with a key created below.
Add the mcpServers entry to claude_desktop_config.json
(Claude Desktop → Settings → Developer → Edit Config) or to a project's
.mcp.json for Claude Code:
Copy this token now — you won't see it again.
For claude_desktop_config.json (Claude Desktop) or .mcp.json (Claude Code):
| Name | Prefix | Created by | Created | Last used |
|---|
No company keys yet. Create one above.