Prototyp · Vorschau: alle Server aus der offiziellen Registry, alle 30 Minuten synchronisiert. Bewertungen und Benutzerkonten folgen in einem späteren Schritt.
PopNx MCPMCP-Server für alle, die KI nutzen · in einfachen Worten erklärt

DataToRAG

von datatoragVersion 1.0.1i★ 4iLäuft in der Cloud
Name in der Registrycom.datatorag/mcp-gateway

Installieren

Überblick

Beschreibung vom Entwickler (englisches Original)

Hosted MCP for Google Workspace, Atlassian and Notion: one OAuth, per-user accounts, write access.

Die drei Säulen von PopNx MCP

Diese drei werden nie zu einer Punktzahl zusammengefasst: Meinungen, Messungen und Beliebtheit sind verschiedene Dinge.

Was du damit machen kannst

Für diesen Server haben wir noch keine Zusammenfassung in einfacher Sprache geschrieben. Unten stehen die Fakten, die wir aus der Registry gelesen haben.

  • Verbindung: Läuft in der Cloud
  • Transport: Streamable HTTP
  • Kategorie: Dokumente, Tabellen & Büroarbeit

Installation

Wähle die KI-App, die du nutzt, und kopiere die Einstellungen.

Bevor du installierst

  • Basischecks: 4 von 4 bestanden
  • Läuft auf dem Server des Entwicklers: Was du fragst, läuft über dessen Server.
  • Kann 60 Dinge (Tools) atlassian-mcp__confluence_add_comment, atlassian-mcp__confluence_create_page, atlassian-mcp__confluence_delete_page, atlassian-mcp__confluence_edit_page …
  • Die Registry nennt keinen Schlüssel zum Eintragen.

Du brauchst

Welche App nutzt du?

Wähle oben deine App. Wir merken sie uns, damit der Button beim nächsten Mal „In … installieren“ anzeigt.

Funktioniert in Claude im Web, auf dem Desktop und auf dem Handy:

  1. Öffne Settings → Connectors
  2. Tippe auf Add custom connector
  3. Füge die Adresse unten ein und tippe auf Add
  4. Wenn eine Anmeldung verlangt wird, melde dich beim Dienst des Entwicklers an
SettingsConnectors1+ Add2https://…3Connect4
https://datatorag.com/mcp

Führe das in deinem Terminal aus

claude mcp add --transport http gateway https://datatorag.com/mcp

In Cursor öffnen

Oder kopier die Einstellungen selbst

In .cursor/mcp.json in deinem Projekt oder in ~/.cursor/mcp.json einfügen

{
  "mcpServers": {
    "gateway": {
      "url": "https://datatorag.com/mcp"
    }
  }
}

Dein Browser fragt, ob er die App öffnen darf, dann zeigt die App die Einstellungen und bittet dich um Bestätigung. Schlüsselwerte bleiben leer, damit du sie selbst einträgst.

In VS Code öffnen

Oder kopier die Einstellungen selbst

In .vscode/mcp.json in deinem Projekt einfügen

{
  "servers": {
    "gateway": {
      "type": "http",
      "url": "https://datatorag.com/mcp"
    }
  }
}

Dein Browser fragt, ob er die App öffnen darf, dann zeigt die App die Einstellungen und bittet dich um Bestätigung. Schlüsselwerte bleiben leer, damit du sie selbst einträgst.

ChatGPT verbindet sich nur mit Servern im Internet:

  1. Öffne ChatGPT im Web → Settings → Connectors (Apps)
  2. Tippe auf Create oder Add (bei manchen Tarifen musst du zuerst Developer mode unter Advanced einschalten)
  3. Füge die Adresse unten ein, gib ihr einen Namen und speichere
  4. Wenn eine Anmeldung verlangt wird, melde dich beim Dienst des Entwicklers an
SettingsConnectors1+ Create2https://…3Connect4
https://datatorag.com/mcp

Menünamen können sich je nach Tarif und mit der Zeit leicht unterscheiden.

Deine Auswahl bleibt in einem Cookie auf diesem Gerät; kein Konto nötig.

Diese Einstellungen werden aus dem Registry-Eintrag des Entwicklers erzeugt. Wenn in der README etwas anderes steht, folge der README.

Welche Apps ihn nutzen können

Basierend darauf, wie sich dieser Server verbindet, und auf der eigenen Doku jeder App, geprüft am 10. Nov. 2026.

Verbinden sich über das Internet in

ClaudeChatGPTGooseLM StudioMsty StudioCherry StudioCursorVS Code + GitHub CopilotVisual StudioWindsurf (Devin Desktop)ZedKiroClineContinueJetBrains AI AssistantClaude CodeCodex CLIGemini CLIOpenCodenodecodeWarpLibreChatOpen WebUIRaycast

Alle Apps, mit Vor- und Nachteilen →

Was er kann

Tool-Namen und -Beschreibungen genau so, wie der Server sie am 10. Nov. 2026 gemeldet hat. Geschrieben hat sie der Entwickler, nicht wir; wir zeigen sie als reinen Text.

  • atlassian-mcp__confluence_add_comment

    Add a comment to a Confluence page: a footer comment on the page, an inline comment anchored to a piece of the page's text with inline_text, or a reply to an existing footer or inline comment with parent_comment_id. A reply goes into the same kind of thread as the comment it answ

  • atlassian-mcp__confluence_create_page

    Create a new page in Confluence. Content must be in XHTML storage format.

  • atlassian-mcp__confluence_delete_page

    Delete a Confluence page by ID.

  • atlassian-mcp__confluence_edit_page

    Update an existing Confluence page. Content must be in XHTML storage format. If version is not provided the current version is fetched and auto-incremented.

  • atlassian-mcp__confluence_get_attachment

    Get metadata for a specific attachment on a Confluence page by filename.

  • atlassian-mcp__confluence_get_comments

    Get all comments on a Confluence page, footer and inline, with their replies, including their body content and version info. Each comment carries its type (footer or inline) and parent_id, the comment it replies to, or null for a comment on the page itself. An inline comment also

  • atlassian-mcp__confluence_get_page

    Get a single Confluence page by ID. Use format 'text' (default) for reading/summarizing — returns clean text with much less context usage. Use format 'storage' when you need to edit the page, as it returns the full XHTML storage format needed for write-back.

  • atlassian-mcp__confluence_list_pages

    List pages in a Confluence space. Returns an array of pages with id, title, version, and link.

  • atlassian-mcp__confluence_search

    Search Confluence content using CQL (Confluence Query Language). Returns matching pages/content with id, title, version, and link.

  • atlassian-mcp__jira_add_attachment

    Add a file to a Jira issue as an attachment. The file is named by a file reference. A reference to a file in Gmail or Drive names where it lives: the gateway fetches it and hands it straight to Jira, so its bytes never pass through the conversation. A gmail_message reference atta

  • atlassian-mcp__jira_add_comment

    Add a comment to a Jira issue.

  • atlassian-mcp__jira_add_worklog

    Log time spent on a Jira issue. Takes the issue key, the time in Jira's duration format ("1h 30m", "2d", "45m"), and optionally when the work started and a comment. Returns the worklog Jira stored: its ID, author, time spent in Jira's words and in seconds, start time and comment.

  • atlassian-mcp__jira_create_issue

    Create a new Jira issue in the specified project. You can set arbitrary fields at creation via additional_fields (e.g. assignee, priority, labels, components) — required for projects that reject unassigned issues. Returns the created issue key and URL.

  • atlassian-mcp__jira_delete_comment

    Delete a comment from a Jira issue.

  • atlassian-mcp__jira_delete_issue

    Permanently delete a Jira issue. THIS CANNOT BE UNDONE through the API — deleted issues do not go to a trash or archive, and the issue key is not reused, so every link to it breaks. Prefer transitioning the issue to Done or Won't Do, which keeps the history. Deleting an issue tha

  • atlassian-mcp__jira_edit_comment

    Edit an existing comment on a Jira issue.

  • atlassian-mcp__jira_get_attachment

    Get metadata for a Jira attachment by its ID. Returns filename, size, MIME type, and content URL.

  • atlassian-mcp__jira_get_comments

    Get all comments on a Jira issue.

  • atlassian-mcp__jira_get_issue

    Get detailed information about a specific Jira issue by its key (e.g. PROJ-123). Returns summary, status, priority, assignee, reporter, description, labels, dates, comments count, and attachments.

  • atlassian-mcp__jira_get_transitions

    Get the available workflow transitions for a Jira issue. Use the returned transition IDs with jira_transition_issue to move an issue through its workflow.

  • atlassian-mcp__jira_link_issues

    Link two Jira issues, such as one blocking another or one duplicating another. Takes the two issue keys and a link type, by its name ("Blocks") or by the phrase it reads with from the first issue ("blocks", "is blocked by", "relates to"), so issue_key blocks other_issue_key, or i

  • atlassian-mcp__jira_list_fields

    List all available Jira fields (both system and custom). Useful for discovering field IDs needed for creating or updating issues.

  • atlassian-mcp__jira_list_issue_links

    List the links of a Jira issue: for each, its ID, type, the phrase that reads from this issue to the other one ("blocks", "is blocked by"), and the other issue's key, summary and status. Also returns the link types the site has, each with its name and its two phrases, which are w

  • atlassian-mcp__jira_list_worklogs

    List the time logged on a Jira issue. Returns each worklog's ID, author, time spent in Jira's words and in seconds, start time and comment as plain text, with the number of worklogs and the total time in seconds.

  • atlassian-mcp__jira_search

    Search Jira issues using JQL (Jira Query Language). Returns matching issues with key fields. Supports pagination via next_page_token.

  • atlassian-mcp__jira_search_users

    Search for Jira users by name, username, or email address. Returns matching user accounts with display names and account IDs.

  • atlassian-mcp__jira_transition_issue

    Transition a Jira issue to a new workflow status. Use jira_get_transitions first to find valid transition IDs.

  • atlassian-mcp__jira_update_issue

    Update an existing Jira issue. You can change the summary, description, and/or set arbitrary fields via additional_fields.

  • gws-mcp__calendar_create_event

    Create a new calendar event. Supports setting title, time, attendees, description, location, and Google Meet links.

  • gws-mcp__calendar_delete_event

    Delete a calendar event by its event ID.

  • gws-mcp__calendar_freebusy

    Check availability (free/busy) for one or more people over a time range. Useful for finding open slots to schedule meetings.

  • gws-mcp__calendar_get_event

    Get details of a specific calendar event by its event ID. Returns the full event (all attendees, reminders, conference data), with the description converted to plain text; use full for the original (often HTML) description.

  • gws-mcp__calendar_list_calendars

    Lists the calendars on the connected account's calendar list: its own, shared ones it has added, and subscribed ones such as holidays. Returns one compact entry per calendar: its ID, title, whether it is the primary calendar, the account's access role (owner, writer, reader or fr

  • gws-mcp__calendar_list_events

    List upcoming events from a Google Calendar. By default returns a compact view per event: id, title, start/end, location, plain-text description (HTML stripped, truncated), organizer email, attendee count plus your own response status, the full attendee roster when the meeting ha

  • gws-mcp__calendar_respond_to_event

    Answers an invitation on the connected account's calendar: accepted, declined or tentative, with an optional note to the organizer. Takes the event ID and the response. The account has to be on the event's guest list; an event it is not invited to is refused with a message saying

  • gws-mcp__calendar_suggest_time

    Suggests meeting times when everyone named is free. Takes the attendees' addresses, a duration and a time range, and reads their free/busy along with the connected account's own. Returns candidate start and end times inside working hours, in order of time, each as an ISO time and

  • gws-mcp__calendar_update_event

    Update an existing calendar event. Only provided fields are changed.

  • gws-mcp__contacts_create

    Create a new contact in Google Contacts.

  • gws-mcp__contacts_delete

    Delete a contact from Google Contacts.

  • gws-mcp__contacts_directory_search

    Search the company's Google Workspace directory (all users in the organization). Useful for finding colleagues' contact info.

  • gws-mcp__contacts_get

    Get full details of a specific contact by their resource name.

  • gws-mcp__contacts_list

    List contacts from the user's Google Contacts. Returns names, emails, phone numbers, and organizations.

  • gws-mcp__contacts_search

    Search Google Contacts by name, email, or phone number. Returns matching contacts with their details.

  • gws-mcp__contacts_update

    Update an existing contact. Only provided fields are changed.

  • gws-mcp__docs_batch_update

    Apply batch updates to a Google Doc. Supports inserting text, replacing text, deleting content ranges, and other document modifications. Uses the Google Docs API batchUpdate format.

  • gws-mcp__docs_create

    Create a new Google Doc.

  • gws-mcp__docs_delete

    Delete a Google Doc. This permanently removes the document from Drive.

  • gws-mcp__docs_get

    Get the content of a Google Doc. Three modes: "text" (default) returns plain text — use for reading/summarizing. "index" returns text with startIndex/endIndex — use before positional edits (insertText at index, deleteContentRange). "full" returns the raw API response — use only f

  • gws-mcp__docs_write

    Insert text at the beginning of a Google Doc. To append or edit at a specific position, use docs_get (mode 'index') then docs_batch_update.

  • gws-mcp__drive_copy_file

    Copies a file in Google Drive and names the copy in the same call. The copy carries the original's tabs, formatting and formulas, so a copied template matches its source. With convert set to true, an Excel, Word or PowerPoint file is copied as a Google Sheet, Doc or Slides file,

  • gws-mcp__drive_create_folder

    Create a new folder in Google Drive.

  • gws-mcp__drive_get_file

    Describes one file or folder in Google Drive without reading its content. Takes the file ID. Returns its name, MIME type, size in bytes when Drive reports one (folders have none), the IDs of the folders it is in, its owners' names and addresses, when it was created and last modif

  • gws-mcp__drive_get_file_permissions

    Lists who has access to a Google Drive file or folder and at what role. Takes the file ID. Returns one entry per permission: its ID, its type (user, group, domain or anyone), the role (owner, organizer, fileOrganizer, writer, commenter or reader), and the person's name and addres

  • gws-mcp__drive_list_recent_files

    Lists the Google Drive files the account opened most recently, or the ones modified most recently, the most recent at the top. Returns one compact entry per file: its ID, name, MIME type, when it was last modified, when the account last opened it, and its link, plus a page token

  • gws-mcp__drive_move_file

    Moves a file or folder in Google Drive into another folder. Takes the file ID and the destination folder ID. The item is added to the destination folder and removed from the folder or folders it was in, so it ends up in one place; its name, content and ID are unchanged. Returns t

  • gws-mcp__drive_read_file

    Read the text content of any file in Google Drive by file ID. Supports Google Docs, Sheets, Slides, Office formats (.docx/.xlsx/.pptx — auto-converted), and plain text files. Returns extracted text directly — no local filesystem needed.

  • gws-mcp__drive_rename_file

    Rename a file or folder in Google Drive. Changes the name only: content, location, and sharing are untouched. Works on any Drive item the account can edit, including Docs, Sheets, Slides, and folders.

  • gws-mcp__drive_search

    Search for files in Google Drive. Returns file names, IDs, types, and modification dates.

  • gws-mcp__drive_share_file

    Shares a Google Drive file or folder with one person by email address, as reader, commenter or writer. Takes the file ID, the address and the role, and optionally a message for the notification email. Google emails the person unless notify is false. Returns the file ID and its pe

  • gws-mcp__drive_trash_file

    Moves a file or folder in Google Drive to Trash. Takes the file ID. The item stays in Trash, where Drive deletes it for good after 30 days and where its owner can restore it before then; trashing a folder trashes what is inside it. Returns the item's ID, name and its trashed stat

Basischecksi

Wir prüfen Informationen, die jeder sehen kann; den Code haben wir nicht gelesen. Checks, die nicht zutreffen, werden nicht mitgezählt. · Geprüft am 10. Nov. 2026

  • ✓ Bestanden

    Entwickler passt zum Repo

    Der Registry-Name wurde über die Domain datatorag.com bestätigt (die Registry prüft, wem die Domain gehört)

  • ✓ Bestanden

    Wird noch gepflegt

    Code zuletzt geändert: heute

  • … Noch nicht geprüft

    Hat eine Lizenz

    Dieses Repo haben wir noch nicht von GitHub gelesen

  • ✓ Bestanden

    Endpunkt antwortet

    Antwort in 135 ms (HTTP 200)

  • – Nicht zutreffend

    Anmeldung mit OAuth

    Keine Anmeldung nötig

  • ✓ Bestanden

    Keine versteckten Anweisungen in Tool-Beschreibungen

    Wir haben 60 Tool-Beschreibungen gelesen und keine versteckten Anweisungen an die KI gefunden

Im Vergleich mit anderen in Dokumente, Tabellen & Büroarbeiti

  • Mehr Sterne als 88 % der anderen 3691 Server in dieser Kategorie
  • Neuer in der Registry veröffentlicht als 91 % der anderen 3691 Server in dieser Kategorie
  • Basischecks gleich gut wie oder besser als bei 100 % der anderen 3691 Server in dieser Kategorie

Verglichen mit allen aktiven Servern dieser Kategorie in der Registry. Sterne zählen nur, wenn dem Herausgeber das Repo gehört.

Häufige Fragen

Ist DataToRAG sicher?

Er hat 4 von 4 zutreffenden Basischecks bestanden. Den Code haben wir nicht gelesen. Bevor du einen Schlüssel oder Token hinzufügst, sieh nach, welche Zugriffe der Server verlangt, und nutze einen Schlüssel, der nur die Berechtigungen hat, die du brauchst.

Welche Schlüssel oder Codes braucht er?

Die Registry nennt keinen erforderlichen Schlüssel, manche Server bitten dich aber bei der ersten Nutzung, dich anzumelden.

Läuft er auf meinem Computer oder in der Cloud?

Er läuft in der Cloud des Entwicklers. Du verbindest dich übers Internet, ohne etwas zu installieren, aber was du sendest, läuft über die Server des Entwicklers.

Wird er noch gepflegt?

Das GitHub-Repo wurde zuletzt heute geändert, und die neueste Version (1.0.1) kam vorgestern in die Registry.

Welche Apps können ihn nutzen?

Apps, die MCP unterstützen, etwa Claude Desktop, Claude Code, Cursor und VS Code. Er verbindet sich per URL, also funktionieren auch Apps, die Remote-MCP-Server akzeptieren, zum Beispiel ChatGPT im Entwicklermodus.

Was kann ich stattdessen nutzen?

In derselben Kategorie: Puter MCP, Agent-Native Slides, Agent-Native Forms

Alternativen

Server derselben Kategorie von anderen Entwicklern, sortiert nach Sternen.

So liest du eine Karte:↓ Downloads pro Wochei★ GitHub-SterneiDrei Säulen: Bewertungen · Checks · NutzungiAktualisiert = neueste Version in der Registry veröffentlichti

Puter MCP

Puter MCP enables AI tools to interact with Puter: manage files, websites, workers, and more

von HeyPuterWeit verbreitet★ 44.122Aktualisiert vor 5 Monaten

Läuft in der Cloud✓ puter.com

Agent-Native Slides

Agent-Native Google Slides - generate and edit React presentations

von BuilderIOWeit verbreitet★ 7665Aktualisiert vor 2 Monaten

Läuft in der Cloud

Agent-Native Forms

Agent-Native form builder - create, edit, and manage forms

von BuilderIOWeit verbreitet★ 7665Aktualisiert vor 2 Monaten

Läuft in der Cloud

README vom Entwickler

README öffnen

Unverändert aus dem Repo des Entwicklers übernommen. PopNx MCP hat den Inhalt weder geprüft noch empfiehlt es ihn. HTML, Bilder und Skripte wurden entfernt. Links führen von dieser Website weg.

Google Workspace, Jira, Confluence and Notion for Claude and other MCP clients, through one endpoint.

DataToRAG is an open-source Model Context Protocol gateway. You point your AI client at one URL. Each person connects their own Google, Atlassian and Notion accounts, and the gateway hands the client the tools for those accounts, calling the services with the person's own OAuth grant. It runs hosted at datatorag.com, or on your own infrastructure under the MIT license.

Website · Docs · Status · Changelog · Skills

Try it in 60 seconds

Setup is two connections, and both are needed:

  1. Connect your accounts. Sign in at datatorag.com/dashboard and connect Google, Atlassian or Notion. Google's consent screen arrives with every box unticked: tick "Select all", or the services you want.
  2. Connect your AI client to https://datatorag.com/mcp, then sign in with the same email when the client asks.

Claude (web and Desktop). Customize, then Connectors, then Add, then "Add custom connector". Paste the URL and leave the advanced OAuth fields empty.

Claude Code.

claude mcp add --transport http datatorag https://datatorag.com/mcp

Cursor and other clients that take a JSON config.

{

"mcpServers": { "datatorag": { "url": "https://datatorag.com/mcp" } } }

ChatGPT. Settings, then Connectors, enable Developer mode, then Create, and paste the URL.

The getting started guide has screenshots of each step. Every client connects with OAuth.

The server is listed in the official MCP registry as com.datatorag/mcp-gateway; server.json is the listing.

Screenshots

The agent in the dashboard runs the same tools against your connected accounts, with no client setup.

An upload link takes one file from your computer, up to 25 MB, and answers its size and SHA-256.

The status page shows the latest end-to-end run against production, by service.

Connectors

Tool counts are from the code as of October 2026: 133 tools in the three connectors, and 10 built into the gateway.

ConnectorToolsServicesDocs
Google Workspace (plugins/gws-mcp)87Gmail, Calendar, Drive, Docs, Sheets, Slides, Contacts, Tasksdocs
Atlassian (plugins/atlassian-mcp)28Jira, Confluencedocs
Notion (plugins/notion-mcp)18Pages, databases, commentsdocs
Gateway built-ins10Connected accounts, skills, file upload and download links, a connectivity checkdocs

The three connectors live in this repository under plugins/ and ship inside the gateway's image. A client only sees the tools for services its user has connected.

What makes it different

Several accounts per service. Connect a work and a personal Google account, or two Notion workspaces, under one login, and pick a default. Every connector tool takes an account argument to choose one (how we built it).

Responses sized for a context window. Tools return what the model needs rather than the raw API response. A two-page Google Doc comes back as 2 to 3 KB of text instead of 50 KB or more of JSON (the write-up).

OAuth per user. Each person grants access to their own accounts, and the gateway refreshes the grants. A connector receives only the calling user's access token, never a shared credential.

Files move between services without passing through the chat. A file reference names an email, an attachment or a Drive file, and the gateway carries it to Jira, Drive or an email in one call, answering with its size and SHA-256. Files from your computer arrive through a single-use upload link (file references).

Skills. Ready-made workflows, such as a morning brief or inbox triage, that run in the dashboard, on a schedule, or from any client as MCP prompts (skills, announcement).

A public status page. A test run calls the tools on production the way a user would and publishes what works, by service (status).

Architecture

In text:

  1. Client. Any MCP client sends Streamable HTTP to POST /mcp with a bearer token: an OAuth access token from the gateway's own authorization server, the only credential it accepts.
  2. Gateway (apps/gateway, one Node.js process: Express plus Next.js). It authenticates the bearer, opens or reuses the session, and answers tools/list from the tools table in Postgres, filtered to the services the user has connected. A tools/call is split on __ into the plugin and the tool, the user's token for that service is resolved (and refreshed if expired), and the call is forwarded to the plugin. Every call is metered. The same process serves the dashboard and its agent, the docs, the changelog and the status page.
  3. Built-ins. list_connected_accounts, skills_*, file_upload_link and file_download_link are answered by the gateway itself, with no plugin and no registry row.
  4. Plugins. Each connector is an MCP server under plugins/. The gateway starts each one as a child process under its own non-root account, with an environment of PATH, NODE_ENV and PORT only, and passes the user's token on each call in the X-User-Token header. A crashing plugin is restarted up to three times a minute, and /health reports each plugin as up or down.
  5. Registry. Postgres holds users, connected accounts, provider tokens and the tool registry (name, description and input schema for each plugin tool). Changes to the registry in production are SQL files under registry/, applied by the registry workflow after an approval.

Self-hosting

Prerequisites
  • Node.js 20 or later and pnpm 9.
  • A Postgres database, and its connection string as DATABASE_URL. Any Postgres works; nothing here starts one for you.
  • An OAuth client for each provider you want: Google (one client for sign-in, a second for Workspace scopes), Atlassian, Notion.
Run the gateway
pnpm install

pnpm -r --filter "./packages/*" run build # the gateway reads these packages' built output

cp .env.example .env # then fill in DATABASE_URL, CONNECTION_TOKEN_KEYS and the OAuth clients pnpm --filter @datatorag-mcp/db db:migrate # schema changes are never applied on start

pnpm dev:gateway

The gateway listens on http://localhost:8285 (set by GATEWAY_PORT):

PathWhat it is
/mcpThe MCP endpoint
/dashboardSign in, connect accounts, the agent
/docs/getting-startedThe docs
/healthLiveness, and each plugin's state
/.well-known/oauth-authorization-serverOAuth metadata for MCP clients

On a laptop the gateway loads plugins from ~/.datatorag/plugins, or from DATATORAG_PLUGINS_DIR when it is set. Build them with pnpm --filter @datatorag-mcp/gws-mcp run build (and the same for atlassian-mcp and notion-mcp).

Environment variables

The full list, with comments, is in .env.example, and the schema that validates it is packages/config/src/index.ts. The ones a self-host needs:

VariablePurpose
DATABASE_URLPostgres connection string. Required
CONNECTION_TOKEN_KEYSKeys the provider tokens are encrypted with: : , comma separated, the first encrypts and all decrypt. Required; make one with openssl rand -base64 32
GATEWAY_PORT, GATEWAY_BASE_URLPort, and the public URL OAuth redirects are built from
GOOGLE_CLIENT_ID, GOOGLE_CLIENT_SECRETGoogle sign-in for the dashboard
GOOGLE_GWS_CLIENT_ID, GOOGLE_GWS_CLIENT_SECRETGoogle Workspace connections
ATLASSIAN_CLIENT_ID, ATLASSIAN_CLIENT_SECRETJira and Confluence connections
NOTION_OAUTH_CLIENT_ID, NOTION_OAUTH_CLIENT_SECRETNotion connections
SLACK_CONNECTOR_CLIENT_ID, SLACK_CONNECTOR_CLIENT_SECRET, SLACK_CONNECTOR_TEAM_IDSlack connections, for one workspace and admins only; optional
ANTHROPIC_API_KEY, PLAYGROUND_MODELThe dashboard agent; empty turns it off
STRIPE_*Billing; optional
POSTHOG_API_KEY, SLACK_BOT_TOKEN, BREVO_API_KEYAnalytics, notifications and email; each is off when empty
Production

Production runs the image built from apps/gateway/Dockerfile, which carries the three plugins. A merge to main builds and publishes it (build.yml), and a deploy is dispatched by hand and approved (release.yml), starting it with docker/docker-compose.prod.yml.

Project structure

apps/gateway/          The product: MCP gateway, OAuth server, dashboard, site

server.ts Express entry: /mcp, /oauth, /auth, /files, /health src/gateway/ Tool routing, auth, plugin manager, file references, metering src/app/ Next.js app: dashb

Nur der Anfang wird gezeigt; den Rest findest du im Repo