MCP YouTube Transcript
An MCP server retrieving transcripts of YouTube videos
Beschreibung vom Entwickler (englisches Original)
Find LinkedIn prospects, draft outreach, and run human-paced campaigns from your AI agent.
Diese drei werden nie zu einer Punktzahl zusammengefasst: Meinungen, Messungen und Beliebtheit sind verschiedene Dinge.
Für diesen Server haben wir noch keine Zusammenfassung in einfacher Sprache geschrieben. Unten stehen die Fakten, die wir aus der Registry gelesen haben.
Wähle die KI-App, die du nutzt, und kopiere die Einstellungen.
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:
https://omentir.com/api/agent/v1/mcpFühre das in deinem Terminal aus
claude mcp add --transport http omentir https://omentir.com/api/agent/v1/mcpIn .cursor/mcp.json in deinem Projekt oder in ~/.cursor/mcp.json einfügen
{
"mcpServers": {
"omentir": {
"url": "https://omentir.com/api/agent/v1/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 .vscode/mcp.json in deinem Projekt einfügen
{
"servers": {
"omentir": {
"type": "http",
"url": "https://omentir.com/api/agent/v1/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:
https://omentir.com/api/agent/v1/mcpMenünamen 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.
Der Server verlangt eine Anmeldung, bevor er seine Tools auflistet.
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
Der Registry-Name wurde mit dem GitHub-Konto vanshyadav1408 bestätigt, dem das Repo gehört
Dieses Repo haben wir noch nicht von GitHub gelesen
Dieses Repo haben wir noch nicht von GitHub gelesen
Antwort in 431 ms (HTTP 401)
Gibt OAuth-Anmeldung an, damit deine KI-App dich zum Einloggen weiterleiten kann
Noch nicht geprüft; wir lesen die Tool-Liste, wenn wir den Endpunkt prüfen
Verglichen mit allen aktiven Servern dieser Kategorie in der Registry. Sterne zählen nur, wenn dem Herausgeber das Repo gehört.
Er hat 3 von 3 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.
Die Registry nennt keinen erforderlichen Schlüssel, manche Server bitten dich aber bei der ersten Nutzung, dich anzumelden.
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.
Das GitHub-Repo wurde zuletzt – geändert, und die neueste Version (1.1.0) kam vor 43 Tagen in die Registry.
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.
In derselben Kategorie: MCP YouTube Transcript, Google Analytics MCP, BibiGPT YouTube Video Summarizer
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
An MCP server retrieving transcripts of YouTube videos
GA4 MCP server for AI agents: schema discovery, server-side aggregation, analysis-ready defaults.
AI video summarizer: summarize and transcribe YouTube, Bilibili, TikTok and podcast videos.
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.
Omentir is an open-source alternative to LinkedIn outreach and automation tools like HeyReach and Gojiberry.
It's an AI sales workspace for finding qualified LinkedIn prospects, drafting contextual outreach, running human-paced campaigns, and handling replies — but you own the code and run it on your own infrastructure.
Use the managed product at omentir.com, or self-host the same application code with your own Firebase, Gemini, and Unipile accounts.
One codebase: hosted cloud and self-host share this repo. RUN_LOCALLY=TRUE switches auth, billing, marketing, and hosted-only mail off. Public brand contacts for omentir.com stay in source on purpose (see src/lib/hosted-identity.ts); they are not used for self-hosted operation.
People use Omentir from chat and coding agents without sharing LinkedIn passwords:
https://omentir.com/api/agent/v1/mcp → sign in and approve Connect workspace → enable tools in the chat. No API key.Authorization: Bearer to the MCP endpoint or REST /api/agent/v1/*.mode: steal_customers + competitor company URLs). If you have more than one company, ask it to list workspaces and switch.Docs: MCP integration, Agent API, agents.md (machine guide), OpenAPI.
Omentir is self-hostable with external managed services. It is not offline or dependency-free. You provide Firebase/Firestore, Unipile, and either a Gemini API key or a Google Cloud Vertex AI project.
Self-hosting removes the Omentir subscription, but Firebase/Google AI and Unipile can still cost money.
Install Git and either Docker with Docker Compose, or Bun 1.3. Then clone and configure the project:
git clone https://github.com/vanshyadav1408/Omentir.git
cd Omentir cp .env.example .env
Open .env and fill in every uncommented blank value. Complete the provider setup below before starting the server.
Generate independent secrets with:
openssl rand -base64 48
Use that (or a password manager) for at least:
LOCAL_SESSION_SECRET (required)LOCAL_APP_PASSWORD (required unless you explicitly allow open access — see Access control)CRON_SECRET (required unless automation is fully disabled)UNIPILE_WEBHOOK_SECRET (required for reply and relation webhooks)For an initial setup with all automation disabled, leave CRON_SECRET blank and set AUTOMATION_DISABLED=true instead. Keep ENABLE_LIVE_AUTOMATION=false until dry-run output has been reviewed.
Docker is the supported self-hosting path. Omentir runs on amd64 and arm64.
docker compose up --build -d
docker compose logs -f omentir
Open http://localhost:3000. Press Ctrl+C to exit the log view without stopping Omentir. Stop the server later with docker compose down.
Port binding: Compose publishes the app as 127.0.0.1:3000:3000 only. That means the container is reachable from the same machine (http://localhost:3000), not from other devices on your LAN or the public internet. This is intentional: a misconfigured or passwordless instance must not become reachable just because Docker started.
To expose the app on a real host (LAN/VPS):
127.0.0.1:3000 (or change the Compose bind only if you understand the risk).APP_BASE_URL to the exact public HTTPS origin (no trailing slash).LOCAL_APP_PASSWORD (do not enable open access on a public URL).For local development:
bun install
bun run dev
For a production-style local server:
bun install
bun run build bun start
Open http://localhost:3000. Sign in with LOCAL_APP_PASSWORD unless you opted into open access (see below).
Bun is the package manager and script runner, but bun run dev deliberately starts Next on Node rather than under bun --bun. In dev, Turbopack loads every package in serverExternalPackages (firebase-admin among them) through a hashed specifier such as firebase-admin- /firestore, which only resolves via a loader hook Next installs in Node. Bun's resolver does not run that hook, so bun --bun next dev fails on the first Firestore import with ResolveMessage: Cannot find module. Production is unaffected: built output emits plain requires, so build and start still run under Bun.
Self-host mode uses a single workspace and a signed, HTTP-only session cookie (omentir_local_session), not Clerk.
| Setting | Behavior |
|---|---|
LOCAL_APP_PASSWORD set (12+ characters, high entropy) | Login form requires that password. This is the default expectation. |
LOCAL_APP_PASSWORD blank and LOCAL_ALLOW_OPEN_ACCESS unset/false | Startup fails. Blank password is no longer enough to run an unprotected instance. |
LOCAL_ALLOW_OPEN_ACCESS=true and blank password | Explicit opt-in: welcome screen with a “Continue to overview” button (no password). Use only on a trusted machine / private network. |
LOCAL_ALLOW_OPEN_ACCESS=true and password set | Password is still required (open-access flag does not weaken a configured password). |
LOCAL_SESSION_SECRET | Required always. Minimum 32 characters, high entropy. Used to HMAC-sign session tokens. Changing it invalidates all existing sessions. |
Additional login hardening that matters for operators:
POST /api/local-auth/login only accepts requests whose Origin matches APP_BASE_URL.?next= return path is sanitized so values like //evil.com cannot open-redirect the browser after sign-in. Only same-origin relative paths (starting with a single /) are allowed.HttpOnly, SameSite=Lax, and Secure when APP_BASE_URL is HTTPS.firebase deploy --only firestore:indexes, or create the indexes from Firestore error links. The source is firestore.indexes.json.firebase deploy --only firestore:rules. Omentir reaches Firestore only through the server-side Admin SDK, so the shipped firestore.rules denies all direct browser/client access. Keep it deployed — it stops anyone from reading or writing your data with a leaked project ID or web API key.GOOGLE_CLOUD_PROJECT, GOOGLE_CLOUD_LOCATION, and GOOGLE_APPLICATION_CREDENTIALS_JSON, enable Vertex AI, and grant the service account Vertex AI User.Copy .env.example to .env. Required local settings are validated at server startup:
| Variable | Scope | Secret | Runtime behavior |
|---|---|---|---|
RUN_LOCALLY=TRUE | local | no | Selects built-in single-workspace authentication |
APP_BASE_URL | shared | no | Canonical server URL; HTTPS required off localhost |
LOCAL_APP_PASSWORD | local | yes* | Required access password unless open access is explicitly allowed; minimum 12 characters when set |
LOCAL_ALLOW_OPEN_ACCESS | local | no | Must be exactly true to allow a passwordless welcome screen |
LOCAL_SESSION_SECRET | local | yes | Independent random value, minimum 32 characters |
LOCAL_USER_NAME / LOCAL_USER_EMAIL | local | no | Display identity for the single local workspace |
FIREBASE_PROJECT_ID | shared | no | Dedicated Firebase project |
FIREBASE_SERVICE_ACCOUNT_KEY | shared | yes | Service-account JSON; malformed JSON fails startup |
UNIPILE_DSN / UNIPILE_API_KEY | shared | mixed | Unipile endpoint and credential |
UNIPILE_WEBHOOK_SECRET | shared | yes | Authenticates reply and relation webhooks |
GEMINI_API_KEY | local/simple | yes | Gemini Developer API path |
| Vertex variables | shared/advanced | yes | Alternative to GEMINI_API_KEY |
CRON_SECRET | shared | yes | Required unless automation is disabled; used by the built-in scheduler and job routes |
ENABLE_LIVE_AUTOMATION | local | no | Live sends remain off unless exactly true |
AUTOMATION_DISABLED | shared | no | Emergency stop; skips scheduler and job execution |
RESEND_API_KEY / RESEND_FROM_EMAIL | optional | yes/mixed | Transactional notifications only in local mode |
ALLOWED_DEV_ORIGINS | dev only | no | Extra hostnames for Next.js HMR when using tunnels/proxies |
\* Treat LOCAL_APP_PASSWORD as a secret even though it is typed by a human.
Generate secrets with a cryptographically secure password manager or openssl rand -base64 48. Placeholder strings such as replace-me are rejected at startup.
http://localhost is supported for local browser use. Any LAN, VPS, tunnel, or internet deployment must use HTTPS. Put a reverse proxy such as Caddy or nginx in front of port 3000 (prefer proxying to the loopback bind above) and set APP_BASE_URL to the exact public origin. Do not expose port 3000 on 0.0.0.0 without TLS and a password.
Unipile connect and reply callbacks require a reachable HTTPS URL that matches APP_BASE_URL.
When a user connects LinkedIn, Omentir registers a server-to-server notify URL with Unipile:
{APP_BASE_URL}/api/connect/callback
That endpoint requires a high-entropy, one-time connect token issued when the flow starts and consumed on success. It also verifies that the returned account exists in the configured Unipile account before saving it. Shared secrets are never placed in callback URLs.
Reply / relation webhooks use the separate Unipile webhook route and the same UNIPILE_WEBHOOK_SECRET (header x-omentir-webhook-secret).
Keep ENABLE_LIVE_AUTOMATION=false, connect LinkedIn, and inspect dry-run decisions in automationRuns. Only set it to true after the account, targeting, message drafts, and quotas are correct. AUTOMATION_DISABLED=true is the emergency stop and also permits omitting CRON_SECRET.
The process can run a built-in scheduler that calls GET /api/jobs/automation-tick with CRON_SECRET (header x-cron-secret or Authorization: Bearer …). Job secret comparison is constant-time. Do not put CRON_SECRET in browser-facing URLs or public logs.
Local mode never sends Omentir marketing, founder, contact-form, or mailing-list email. Operational notifications can use your own Resend account. Without it, outreach still works and notification sends are skipped.
The same repository powers oment
Nur der Anfang wird gezeigt; den Rest findest du im Repo