fonio MCP (community)
Allows building and configuring WhatsApp agents through fonio.ai, including paste-ready assistant prompts, knowledge base drafting, and voice/WhatsApp agent creation.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@fonio MCP (community)Search the help center for how to trigger an outbound call."
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Unofficial fonio MCP
Use Claude, ChatGPT, Cursor, or another MCP client to build Fonio assistants, prepare outbound calls, search Fonio guidance, and connect remote integrations.
This is an independent MIT-licensed community project. It is not made by, endorsed by, or affiliated with fonio GmbH. It is not a hosted product or a Fonio login service.
What it can do
Build a complete, paste-ready voice or WhatsApp assistant.
Draft the prompt, greeting, knowledge-base Q&A, tools, transfers, calendar, webhooks, and setup checklist.
Validate prompts against the documented Fonio guidance.
Test an API key and trigger outbound calls with an explicit confirmation step.
Register remote integration-manifest servers in Fonio.
Optionally create or update an assistant through the visible Fonio dashboard in a local browser.
Fonio’s public API does not currently provide assistant create/update endpoints. The local browser option exists for that missing dashboard step; it does not use private Fonio HTTP APIs.
Related MCP server: AI Directory MCP Server
Quick start: local MCP
git clone https://github.com/fathyshalaby/fonio-mcp-community.git
cd fonio-mcp-community
npm install
cp .env.example .env.local
npm run mcpFor local API access, set FONIO_API_KEY in .env.local. Create a key in Fonio API keys. Never put your Fonio password in the environment file.
Add the running MCP to your client using its local stdio configuration. For Claude Code:
claude mcp add unofficial-fonio-mcp -- npm run mcpCreate an assistant through the local dashboard
The public API can’t save assistants, so local users can opt into a visible Chrome bridge. It uses a separate browser profile on your machine. You log in manually; credentials and session data never leave the machine.
Add this to .env.local:
FONIO_ENABLE_BROWSER=trueThen run npm run mcp and use the tools in this order:
get_dashboard_browser_statusopen_dashboard_browser, then log in atapp.fonio.aiprepare_dashboard_assistantReview the plan and explicitly confirm it.
Call
create_dashboard_assistantorupdate_dashboard_assistant.
The bridge fills the assistant name, greeting, and prompt when Fonio exposes those fields accessibly. It reports the remaining voice, language, knowledge, calendar, transfer, webhook, and tool settings for review. It will not guess if the Fonio UI changes.
Optional browser settings:
# Use a specific dedicated profile directory
FONIO_BROWSER_PROFILE_DIR=/absolute/path/to/fonio-profile
# Use a Chrome instance started with remote debugging
FONIO_BROWSER_CDP_URL=http://127.0.0.1:9222
# Only if Chrome is not found automatically
FONIO_BROWSER_EXECUTABLE_PATH=/absolute/path/to/chromeNever enable the browser bridge on a hosted deployment.
Connect a hosted MCP
You can self-host the HTTP MCP on your own HTTPS origin and add:
https://your-host.example/mcpHosted mode uses the community OAuth wrapper and a Fonio workspace API key. The local browser bridge is intentionally unavailable in hosted mode.
The repository has no product frontend. The only browser page retained is the small OAuth consent screen required to authorize an HTTP MCP client.
Vercel
npx vercel --prodSet these environment variables:
Variable | Purpose |
| Random secret for OAuth/session encryption |
| Public HTTPS URL of the deployment |
Docker
docker build -t unofficial-fonio-mcp .
docker run --rm -p 43147:43147 \
-e FONIO_MCP_SECRET="$(openssl rand -hex 32)" \
-e NEXT_PUBLIC_MCP_ORIGIN="https://your-host.example" \
unofficial-fonio-mcpMain tools
Tool group | Tools |
Assistant building |
|
Local dashboard |
|
Fonio docs |
|
Outbound calls |
|
Remote integrations |
|
Outbound calls can cost money and require the Fonio Teams plan, KYC, and an imported or SIP number. The tool always requires an exact destination confirmation before dialing.
Development
npm run dev
npm run mcp
npm test
npm run lint
npm run typecheckThe public API reference is app.fonio.ai/api/docs. If this repository disagrees with the official Fonio app or docs, trust the official source.
License and disclaimer
MIT licensed and provided “as is”. This project stores no Fonio passwords and does not claim to be an official Fonio integration. You are responsible for your own deployment, API keys, browser profile, calls, costs, and data.
This server cannot be deployed
Maintenance
Related MCP Connectors
Create, configure and manage an AI phone receptionist. Requires a Fonixia account.
Verified, pay-per-use API tools for AI agents through one authenticated connection.
- DialMCPOAuthcom.dialmcp
Let AI agents place real phone calls from your verified number, with transcripts and recordings.
Give your AI agent a phone. Place outbound calls to US businesses to ask, book, or confirm.
Related MCP Servers
- AlicenseNot gradedqualityCmaintenanceEnables AI agents to manage phone numbers, send/receive SMS, and place voice calls through natural language, connecting to the phone network via the AgentPhone API.296 npm123MIT
- FlicenseAqualityFmaintenanceEnables AI assistants to query the Voxie AI Phone Number Directory, list AI services, and obtain webchat URLs to interact with Voxie AI personas.43-
- FlicenseNot gradedqualityDmaintenanceConnects AI assistants to the Fathom meeting recording API to retrieve meeting summaries, transcripts, and action items.-
- FlicenseNot gradedqualityBmaintenanceEnables AI assistants to look up and update WorkForceAI platform data, including voice agents, campaigns, phone numbers, call logs, and dashboard stats, through natural language.-