
ToughTongue AI
com.toughtongueai由 toughtongueai.com 發布官方發布者v1.0.0更新於 Oct 26, 2026
Create voice-agent scenarios, pull session analytics, place SIP calls, schedule meeting bots.
概覽
Tough Tongue AI 的託管 MCP 伺服器,讓助理在工作區中建立語音代理情境、分析練習工作階段、撥打 SIP 電話並安排會議機器人。
- 可以做什麼
- 建立、編輯與調整語音代理情境與角色扮演,並管理工作區與分享。
- 取得工作階段評分與逐字稿,整理成團隊或教練報告與改進重點。
- 撥打 SIP 電話、安排會議機器人,並為產品示範編寫瀏覽器操作步驟。
- 建置 Agent Desktop 應用程式,例如通話中由語音代理更新的看板或清單。
- 適用情境
- 適合想讓程式或網頁助理建置並迭代 Tough Tongue AI 語音代理與角色扮演、檢視練習或真實通話的表現,或為談判、面試等特定對話做準備的情況。
- 執行需求
- 透過 HTTPS 存取的遠端 MCP 伺服器,不需要 API 金鑰或環境變數。首次使用需在瀏覽器完成 OAuth 登入,因此用戶端需能開啟瀏覽器(無介面或 SSH 環境需依文件使用無瀏覽器登入方式)。需要 Tough Tongue AI 帳號與工作區。
安裝
在 SourceWeft 中
- 開啟 儀表板中的 ToughTongue AI,將其新增到工作區。
- 連接 所需的 憑證。它們會被私密儲存,絕不會在公開頁面上展示。
- 為需要使用其工具的對話啟用該服務。
Web executable,透過 Streamable HTTP。 遠端服務在工作區中設定後即可從網頁執行環境執行。
交給代理程式安裝
把這段提示詞貼上到 Claude Code、Codex、Cursor 等代理程式,或 SourceWeft 對話中。代理程式會讀取該伺服器的安裝指南,先向你展示發布者、端點和登入方式,經你同意後再把它新增到你的客戶端。金鑰不會寫進任何命令。
閱讀 https://sourceweft.com/mcp/com.toughtongueai%2Fmcp/install.md,並按其中說明把這個 MCP 伺服器接入我的客戶端。修改任何設定之前,先告訴我它的發布者、端點和登入方式,等我確認。不要把 API 金鑰寫進命令或我會提交的檔案;問我金鑰應從哪裡取得。其他 MCP 客戶端
把它新增到你客戶端的 mcpServers 設定中。 客戶端會在首次連接時完成 OAuth 登入。
{
"mcpServers": {
"toughtongue-ai": {
"type": "http",
"url": "https://api.toughtongueai.com/api/public/mcp"
}
}
}README
Tough Tongue AI Skills
Tough Tongue AI is the platform for tough conversations. Some the AI takes: voice agents that call, demo, screen, and book. Others you nail: realistic roleplay for negotiations, interviews, and coaching. This plugin lets your coding agent build, run, and improve both, in your personal or organization workspace.
How it works
Armed with these skills, your coding agent becomes a workspace commander for your Tough Tongue AI voice agents.
What the plugin contains and connects to
- Skills and configuration only. Markdown skills, reference docs, an icon, and MCP server configs. No hooks, scripts, or executable code.
- One server. Your agent talks to Tough Tongue AI's hosted MCP server at
https://api.toughtongueai.com/api/public/mcpover HTTPS. You sign in with OAuth in your browser. The plugin reads no API keys, tokens, or environment variables, and its configs carry only that URL. - What is sent. Only what your agent passes to the
ttaitools: Scenario text, session queries, and the phone numbers, meeting links, transcripts, or recording links you ask it to use. It all stays in your Tough Tongue AI workspace under the privacy policy.
Get started
1. Install
Coding agents
These get the skills and the MCP server together.
Any coding agent (Claude Code, Cursor, Codex, Copilot CLI, VS Code)
Claude Code
Codex
Cursor
Settings > Plugins > Team Marketplaces > Add Marketplace > Import from Repo,
enter https://github.com/tough-tongue/toughtongue-skills, then install
toughtongue.
Windsurf, Gemini CLI, and other coding agents
Then add a remote (HTTP) MCP server with the URL
https://api.toughtongueai.com/api/public/mcp. Config snippets:
clients.md.
Web agents
These connect to the MCP server directly.
claude.ai and Claude Desktop
Settings > Connectors > Add custom connector. Name it Tough Tongue AI, paste
https://api.toughtongueai.com/api/public/mcp, and approve the sign-in.
ChatGPT
Settings > Apps, find Tough Tongue AI in the app directory, and connect.
2. Sign in
Restart your agent. The first Tough Tongue AI action opens a browser sign-in;
approve it once. No API key needed. Claude Code over SSH: run
claude mcp login plugin:toughtongue:ttai --no-browser in an ssh -t session
and open the printed URL on your own machine. Running in CI or on a server with
no browser? See headless setup
in the Tough Tongue AI docs.
3. Ask
Get me started with Tough Tongue AI.
Then try:
- "Create a practice scenario for an enterprise pricing negotiation."
- "For the Enterprise Discovery Call scenario, pull the last 50 sessions. What are the top 5 improvement areas?"
- "The onboarding agent keeps ending calls too early. Find out why and refine it."
Skills
- ttai-agent: create, edit, and refine Scenarios; workspaces, calls, meeting bots, and sharing.
- ttai-session-analyst: turn session scores and transcripts into team and coaching reports.
- ttai-browser-demo-builder: script browser steps for product demos from a recording or a walkthrough.
- ttai-agent-apps: build Agent Desktop apps, such as a board or checklist the voice agent updates during a call.
Your agent picks the right skill on its own. In Claude Code you can also call
one directly, for example /toughtongue:ttai-agent.
What you can do
- Practice a lost deal. "Pull the last 3 calls from Gong where we lost on pricing and create a scenario to practice that objection."
- See how the team is doing. "Top 5 improvement areas from the last 50 Enterprise Discovery Call sessions, as a 3-slide deck."
- Improve an agent from real calls. "Find what went wrong in the 5 lowest-scoring onboarding sessions and refine the scenario."
- Prep for a meeting. "I meet Sarah Chen from Acme in 30 minutes. Create a quick practice scenario."
- Score real calls automatically. "Every time a call ends in Gong, score the transcript and email the rep a coaching report."
Update
Claude Code
Codex
npx plugins add and Cursor's import always install the latest version.
Troubleshooting
- Sign-in or 401 errors: finish the browser sign-in. Claude Code:
/mcp(over SSH:claude mcp login plugin:toughtongue:ttai --no-browser, connected withssh -t). Codex:codex mcp login ttai. Cursor: Settings > MCP. - A Tough Tongue AI tool is missing: start a new chat; if it persists, remove and re-add the MCP server.
- The agent explains but can't act: the skills are installed but the MCP server is not. Add it for your client (clients.md).
- A Scenario change doesn't show in a running call: edits apply to new sessions.
Links
Platform docs · Privacy policy · Terms of service
License
MIT
來源:README.md,提交 c311a93
工具
0版本歷史
1- v1.0.0最新Sep 16, 2026
評分
已滿足適用檢查項的 85%
該評分衡量此收錄條目的完整程度,依據註冊中心條目、端點與公開來源計算得出。它不是對伺服器本身的評價,也不表示伺服器安全或目前可用:可連通性是帶日期的檢查結果,評分也不包含安裝量或使用者評價。
- 近 30 天內可連通必需25 分
- 已宣告認證方式必需10 分
- 已列出工具15 分
請讓伺服器在無憑證時回應 tools/list,或發布可在沙箱中無憑證啟動的軟體包。
- 描述必需15 分
- 圖示10 分
- 文件15 分
- 授權條款10 分