MCP YouTube Transcript
An MCP server retrieving transcripts of YouTube videos
作者による説明(英語の原文)
A local CRM as Markdown files in git: read and update pipeline, contacts and interactions.
この3つを1つのスコアにまとめることはしません。意見、測定、人気はそれぞれ別のものだからです。
このサーバーのやさしい解説はまだ書いていません。以下はレジストリから読み取った事実です。
使っているAIアプリを選んで、設定をコピーしてください。
HERMITCRM_DATA必要なもの
どのアプリを使っていますか?
上でアプリを選んでください。選んだアプリを覚えておくので、次からはボタンに「…にインストール」と表示されます。
Claude Desktop → Settings → Developer → Edit Config を開き、claude_desktop_config.json にこれを貼り付けます
{
"mcpServers": {
"hermit-crm": {
"command": "uvx",
"args": [
"hermitcrm",
"mcp"
],
"env": {
"HERMITCRM_DATA": "your-value"
}
}
}
}ターミナルでこれを実行します
claude mcp add hermit-crm --env HERMITCRM_DATA=your-value -- uvx hermitcrm mcp動かす前に、空欄の値を入力してください。
プロジェクトの .cursor/mcp.json、または ~/.cursor/mcp.json に貼り付けます
{
"mcpServers": {
"hermit-crm": {
"command": "uvx",
"args": [
"hermitcrm",
"mcp"
],
"env": {
"HERMITCRM_DATA": "your-value"
}
}
}
}ブラウザがアプリを開いてよいか確認し、そのあとアプリが設定を表示して確認を求めます。キーの値は空欄のままなので、自分で入力してください。
プロジェクトの .vscode/mcp.json に貼り付けます
{
"servers": {
"hermit-crm": {
"type": "stdio",
"command": "uvx",
"args": [
"hermitcrm",
"mcp"
],
"env": {
"HERMITCRM_DATA": "your-value"
}
}
}
}ブラウザがアプリを開いてよいか確認し、そのあとアプリが設定を表示して確認を求めます。キーの値は空欄のままなので、自分で入力してください。
これはパソコンで動くサーバーですが、ChatGPTがつなげるのはインターネット上のサーバーだけです。ほかのアプリを選んでください。
選んだアプリは、この端末のCookieに保存されます。アカウントは不要です。
この設定は、作者のレジストリ登録情報から作成しています。READMEと違う場合は、READMEに従ってください。
パソコンで動くサーバーです。インストールすると、ツール一覧がAIアプリに表示されます。
誰でも見られる情報をチェックしており、コードは読んでいません。当てはまらない項目は数に入れていません。 · 2026年11月6日にチェック
レジストリ名は、リポジトリを所有するGitHubアカウント Hermit-CRM で確認済みです
このリポジトリはまだGitHubから読み取っていません
このリポジトリはまだGitHubから読み取っていません
これはパソコンで動くため、インターネット上のエンドポイントはありません
パソコンで動き、代わりにローカルの設定を使います
パソコンで動くサーバーです。ツール一覧を読み取れるのは、インターネット上のエンドポイントがある場合だけです
レジストリでこのカテゴリにある有効なすべてのサーバーと比較しています。スターは、公開者がリポジトリの所有者である場合だけ数えます。
当てはまる基本チェック1項目のうち1項目に合格しました。コードは読んでいません。キーやトークンを追加する前に、サーバーがどんなアクセスを求めているかを確認し、必要な権限だけを持つキーを使ってください。
入力が必要なもの:HERMITCRM_DATA
あなたのパソコンで動きます。必要なときにAIアプリが起動し、データは作者のサーバーを通りません(サーバー自体がオンラインサービスを呼び出す場合を除きます)。
GitHubリポジトリの最終更新は–、最新バージョン(1.0.0)がレジストリに届いたのは10 日前です。
Claude Desktop、Claude Code、Cursor、VS CodeなどのMCP対応アプリで使えます。パソコンにPython (uv)も必要です。
同じカテゴリには:MCP YouTube Transcript, Google Analytics MCP, BibiGPT YouTube Video Summarizer
同じカテゴリで、ほかの作者が作ったサーバー。スターの多い順。
カードの見方:★ GitHubスターi3つの柱:レビュー · チェック · 利用状況i更新=最新バージョンをレジストリに公開した日i
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.
作者のリポジトリから書かれたとおりにコピーしたものです。PopNx MCPは内容を確認・推奨していません。HTML、画像、スクリプトは取り除いています。リンクはこのサイトの外へ移動します。
A free, local CRM for solopreneurs. Your pipeline, contacts and conversations live as Markdown files in git on your own Mac, with a web app for you and a CLI and MCP server for your AI agent. No account, no cloud, Apache 2.0.
Install it with your AI coding agent. Paste this into Claude Code, Codex, Cursor or any other:
Read https://raw.githubusercontent.com/Hermit-CRM/hermit-crm/main/INSTALL.md and set up Hermit CRM for me.
Prefer to type the commands yourself? See Install.
| Pipeline | A company | Calendar |
|---|---|---|
Alpha software. Hermit CRM is incomplete, it changes without notice, and
it has bugs nobody has found yet. It is free and comes with **no warranty of
any kind** — nobody is on the hook if it loses your data. Keep your own
backups, and check that they restore. Read DISCLAIMER.md
before you put real work in it.
Your CRM is a folder of Markdown files in git. Every company, contact and conversation is a small text file you can read, grep, edit and diff; every change the app makes is a commit, so the previous version of a file stays in git log and most mistakes are one git revert away. A local web app gives you a pipeline board, a calendar of next steps, message drafts and reports, and because the data is plain files with a small CLI, AI agents (Claude Code, Codex and others) can read and update your pipeline as easily as you can. Hermit CRM is built for one person: a founder, a freelancer, someone doing their own sales.
A modest, factual comparison (checked September 2026; corrections welcome):
| Hermit CRM | HubSpot Free | Twenty | Obsidian CRM plugin | |
|---|---|---|---|---|
| Data as Markdown files | Yes | No | No (Postgres) | Yes |
| Git history of every change | Yes, one commit per write | No (activity log) | No | Only with a git plugin |
| Web app | Yes, local | Yes, hosted | Yes | No (inside Obsidian) |
| Email capture | BCC/forward via Gmail IMAP | Yes (BCC, inbox sync) | Yes (Gmail/Outlook sync) | Depends on the plugin |
| AI-agent friendly | Plain files, CLI, AGENTS.md | Via API | Via API | Plain files |
| Self-hosted without Docker | Yes (git clone + pip) | No (SaaS only) | No (Docker Compose) | n/a (desktop app) |
| Multi-user teams | No | Yes | Yes | No |
If you need a team CRM, use one of the others. If you want your pipeline in files you own, Hermit CRM is for you.
With an AI assistant. Paste the line at the top of this page into Claude Code, Codex, Cursor or another AI coding tool, or unpack the download, open the folder in one and send it one message:
Read INSTALL.md and set up Hermit CRM for me.
It checks the machine, installs the command, makes your data folder, asks your name and sending addresses, and starts the app. It will not ask you for an app password or a calendar URL: those you type into the Settings page yourself, so they stay on your machine. INSTALL.md is the script it follows.
Hermit CRM runs on macOS. Linux works but is less tested.
By hand. You need Python 3.11 or newer and git:
git clone https://github.com/Hermit-CRM/hermit-crm.git && cd hermit-crm
python3 -m venv .venv && source .venv/bin/activate pip install . hermitcrm init ~/crm # a new data folder; asks 3 questions hermitcrm --data ~/crm serve # http://127.0.0.1:8765 hermitcrm --data ~/crm schedule install --serve # daily sync, web app always on
On Debian and Ubuntu, python3 -m venv needs the python3-venv package, and Ubuntu 22.04's Python (3.10) is too old; uv (uv tool install . from the clone) brings its own Python and avoids both.
A download is hermitcrm- .tar.gz and unpacks to hermitcrm- /, with a .sha256 beside it. Check it before you unpack:
shasum -a 256 -c hermitcrm-<version>.tar.gz.sha256 # macOS
sha256sum -c hermitcrm- .tar.gz.sha256 # Linux
Hermit CRM is also on PyPI: uv tool install hermitcrm (or pipx install hermitcrm) replaces the first three lines and gives you a hermitcrm command that works without activating the virtualenv.
hermitcrm init asks three questions: you (your name and sending addresses), BCC capture (the tracking address and its app password) and an online copy (an optional private git remote; local backups start with schedule install or Settings > Backup). Skip any of them; rerun with hermitcrm setup or open the Settings page (/settings) in the web app. Add --demo for thirteen fictional companies, --no-setup to skip the questions. Instead of --data you can set HERMITCRM_DATA=~/crm or run commands from inside the folder.
hermitcrm doctor checks the whole installation in one go. Every page of the web app has a Help link; hermitcrm help [topic] prints the same pages.
Want to see a worked account first? The empty home page offers Look at a sample account (or run hermitcrm sample add): one made-up company with its people, messages, a deal on the pipeline and tasks, plus a pinned dashboard and a paused routine to show what you can build. A bar on every page says it is there; Remove it deletes it again in one commit (hermitcrm sample remove), and never touches your own companies.
With the app running at http://127.0.0.1:8765, click New company, type a name and a country, and save. Or do the same from the terminal:
hermitcrm --data ~/crm add company "Acme BV" --country NL --website acme.example.com
Either way one file now exists, and it is the whole record:
~/crm/companies/acme/company.md
name: Acme BV slug: acme website: https://acme.example.com country: NL source: other stage: prospect stage_changed: 2026-09-17 ...
and one commit records it, company: acme created. Open the company page and add a contact, then log the first email:
hermitcrm --data ~/crm add contact acme "Jane Roe" --title CTO --email jane@example.com
hermitcrm --data ~/crm add interaction acme --channel email --direction out \ --contact jane-roe --subject "Intro" --body "Hi Jane, ..."
The contact becomes companies/acme/contacts/jane-roe.md, the mail becomes a file under companies/acme/interactions/, and Acme moves from prospect to engaged because you have now spoken to them. Three files, three commits, git log for the history. hermitcrm show acme prints the lot.
That is the entire data model: a folder per company, a file per contact and a file per conversation. Everything below is convenience on top of it.
hermitcrm followups.hermitcrm brief.messages.toml.outcomes in config.toml)..xlsx.hermitcrm serve --host 0.0.0.0 puts the app on your local network. It has no password, so use a network you trust or a private one (Tailscale, WireGuard).hermitcrm mcp): read and write the CRM from Claude Desktop, ChatGPT, Cursor or anything else that speaks MCP, with no terminal. Ten read tools and seven write tools (records, Make it yours, bulk changes, undo), over stdio, with no extra dependency./help/feedback: writes a report into your folder with the version, counts and feature flags attached, and shows it to you to send. Nothing leaves your machine on its own, and no names or paths go in the report.hermitcrm init writes config.toml with every key commented out at its default. Uncomment what you want to change.
| Key | Default | What it does |
|---|---|---|
owner_name | "" | Your name; its first word signs outreach drafts ({owner_first_name}). |
owner_email | "" | Your main email address (informational). |
| port | 8765 | Port of the web app on 127.0
冒頭だけを表示しています。続きはリポジトリで読めます