> ## Documentation Index
> Fetch the complete documentation index at: https://docs-legacy.famulor.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Famulor Skill

> Install and use the Famulor.io Skill to let AI agents build and manage assistants, campaigns, knowledge bases, and messaging workflows via the Famulor API.

# Famulor.io Skill

The **Famulor.io Skill** enables AI agents to execute real tasks in Famulor, not just explain them.

<div style={{ position: "relative", paddingBottom: "56.25%", height: 0 }}>
  <iframe style={{ position: "absolute", top: 0, left: 0, width: "100%", height: "100%", border: 0 }} src="https://www.tella.tv/video/neuer-famulo-github-skill-automatisierung-per-prompt-hbd4/embed?b=0&title=0&a=1&loop=0&t=0&muted=0&wt=0&o=1" allowFullScreen allowTransparency />
</div>

<Card title="GitHub Repository" icon="github" href="https://github.com/bekservice/Famulor-Skill">
  Source code, installation guides, and examples
</Card>

## What this skill does

* Create and configure AI assistants (inbound, outbound, chat, WhatsApp)
* Set up outbound campaigns and manage leads
* Connect knowledge bases and documents
* Create mid-call tools (HTTP tools) for live integrations
* Run WhatsApp and SMS workflows
* Support testing and iterative prompt and voice optimization

## The guided onboarding flow

The heart of the skill: a structured **4-phase flow** that turns a client conversation into a fully deployed assistant. The agent asks phase by phase — not everything at once.

| Phase                             | Content                                                                                                             |
| --------------------------------- | ------------------------------------------------------------------------------------------------------------------- |
| **1 — Discovery**                 | Company name, industry, assistant name, call direction (inbound/outbound)                                           |
| **2 — Technical configuration**   | Engine type (pipeline / multimodal / dualplex), primary and secondary languages, voice (male/female), ambient sound |
| **3 — Intelligent configuration** | Industry-specific tasks, knowledge base (website/documents), post-call schema, initial greeting                     |
| **4 — System prompt & creation**  | Generate a tailored system prompt, client confirms, assistant is created via API                                    |

Behind it sits **niche intelligence**: a bundled knowledge base (`references/nischen_intelligenz.md`) with industry know-how for hair salons, medical practices, restaurants, real estate, trades, law firms, and more — the agent proactively asks the right follow-up questions per industry.

## All functions

The skill ships with a Python client (`scripts/famulor_client.py`) that calls the Famulor API directly:

| Area                   | Functions                                                                                                 |
| ---------------------- | --------------------------------------------------------------------------------------------------------- |
| **Assistants**         | `create_assistant` · `update_assistant` · `list_assistants`                                               |
| **Configuration data** | `get_languages` · `get_voices` · `get_models` · `get_synthesizer_providers` · `get_transcriber_providers` |
| **Knowledge bases**    | `create_knowledgebase` · `list_knowledgebases` · `create_document` (website import & files)               |
| **Phone numbers**      | `get_phone_numbers` · `list_phone_numbers`                                                                |
| **Webhooks**           | `enable_webhook` (inbound) · `enable_conversation_ended_webhook` (conversation ended)                     |
| **Testing**            | `create_test_conversation` · `send_test_message` — test the assistant via chat before going live          |

### Built-in assistant tools

During creation the skill configures the built-in conversation tools with industry-specific descriptions:

| Tool                       | Purpose                                                                               |
| -------------------------- | ------------------------------------------------------------------------------------- |
| **`end_call`**             | ALWAYS enabled — defines, per niche, when the bot should hang up                      |
| **`call_transfer`**        | Call forwarding to a fallback number (emergencies, urgent cases, request for a human) |
| **`calendar_integration`** | Direct appointment booking via Cal.com / Calendly                                     |

The skill also applies proven defaults: GPT-4.1-mini, ElevenLabs TTS, interruptions and fillers enabled, noise cancellation, call recording, post-call evaluation with a custom schema.

## How the skill is built

The skill follows the [Open Plugins specification](https://open-plugins.com) and consists of four building blocks:

```text theme={null}
skills/famulor-skill/
├── SKILL.md                          # Agent instructions: the 4-phase flow
├── references/
│   └── nischen_intelligenz.md        # Industry knowledge (10+ niche profiles)
├── scripts/
│   └── famulor_client.py             # Python client for the Famulor API
└── templates/
    └── example_template.txt          # System prompt templates
```

| File                         | Role                                                                                                                                                                                                                                                             |
| ---------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **`SKILL.md`**               | Main instruction with frontmatter (`name`, `description` — controls when the agent activates the skill) and the full onboarding flow                                                                                                                             |
| **`nischen_intelligenz.md`** | Profiles for hair salons/beauty, real estate, medical practices, restaurants, car dealers, trades, hotels, gyms, law firms, vets + a fallback for unknown niches — each with typical tasks, proactive questions, post-call schema, and recommended configuration |
| **`famulor_client.py`**      | API client, also usable as a CLI: `python3 famulor_client.py <method> key=value`                                                                                                                                                                                 |
| **`templates/`**             | Reusable system prompt templates (support, appointment booking, sales, lead qualification) with `{{placeholders}}`                                                                                                                                               |

On top of that: manifests for each platform (`.plugin/plugin.json`, `.claude-plugin/`, `.cursor-plugin/`, `gemini-extension.json`, `GEMINI.md`) and the packaged archive `famulor.skill` for manual installation.

## Extending the skill or building your own

The repository is designed as a template — three typical customizations:

**Add a new industry:** Add your own profile to `references/nischen_intelligenz.md` following the existing pattern: typical tasks, proactive questions, recommended configuration (knowledge base yes/no + mode, post-call schema with max 16 characters per field name, ambient sound), system prompt priorities, and an `end_call` description.

**Custom prompt templates:** Drop additional templates into `templates/` and reference them in `SKILL.md` — the agent uses them as a starting point instead of writing from scratch.

**Extend API functions:** Add new methods to `famulor_client.py` (e.g. campaigns or leads) — every public method is automatically callable via the CLI. Endpoints: [API reference](/en/api-reference/).

Then repackage and bump the version:

```bash theme={null}
# Bump version in .plugin/plugin.json, .claude-plugin/, .cursor-plugin/,
# gemini-extension.json, then:
zip -r famulor.skill skills/
```

<Tip>
  For building entirely new skills (different platform, different use case): reuse the structure — `SKILL.md` with a precise `description` (it controls triggering), knowledge in `references/`, executable code in `scripts/`. A good `description` names concrete trigger phrases.
</Tip>

## When to use this skill

Use this skill for requests related to:

* New client onboarding ("new client", "create a bot", "set up a phone bot")
* Famulor or famulor.io setup
* Campaigns, leads, and outbound calling
* WhatsApp bots, templates, and SMS
* Knowledge bases, RAG documents, and webhooks
* Famulor API integrations

**Example prompts:**

```text theme={null}
Onboard a new client: Dr. Meier dental practice in Berlin.
```

```text theme={null}
Create an inbound bot for my restaurant — walk me through the setup.
```

```text theme={null}
Test the assistant "Sofia" via chat: book a table for 4 on Friday.
```

More scenarios: [Use Cases & Prompts](/en/mcp/use-cases).

<Note>
  **Skill vs. MCP server:** The full [MCP server](/en/mcp/server) exposes a broad platform tool catalog (calls, campaigns, conversations, and more) via OAuth. The developer skill focuses on **guided onboarding with industry knowledge** and runs in coding agents (Cursor, Codex, Gemini CLI) via API key. They complement each other. The separate Assistant & History store profile is read-only.
</Note>

## Requirements

You need a valid API key as an environment variable:

```bash theme={null}
export FAMULOR_API_KEY="your-api-key"
```

Create your API key in the Famulor dashboard: [app.famulor.io](https://app.famulor.io).

## Installation

### Cursor

```text theme={null}
/add-plugin https://github.com/bekservice/Famulor-Skill
```

If direct URL installation is not available, import `famulor.skill` manually and restart the chat session.

### Codex

```text theme={null}
Fetch and follow instructions from https://raw.githubusercontent.com/bekservice/Famulor-Skill/refs/heads/main/.codex/INSTALL.md
```

### OpenCode

```text theme={null}
Fetch and follow instructions from https://raw.githubusercontent.com/bekservice/Famulor-Skill/refs/heads/main/.opencode/INSTALL.md
```

### Gemini CLI

```bash theme={null}
gemini extensions install https://github.com/bekservice/Famulor-Skill
```

To update:

```bash theme={null}
gemini extensions update famulor-skill
```

### Universal manual installation

1. Download `famulor.skill` from the repository.
2. Import/register it in your agent platform as a custom skill or plugin.
3. Restart the agent session.
4. Set `FAMULOR_API_KEY`.
5. Ask the agent to execute a Famulor workflow.

## Local developer quickstart

1. Clone the repository.
2. Ensure Python 3.10+ is installed.
3. Set your API key:

```bash theme={null}
export FAMULOR_API_KEY="your-api-key"
```

4. Run:

```bash theme={null}
python3 scripts/famulor_client.py list_assistants
```

If your key is valid, you receive API data as JSON.

## Verify installation

* `echo $FAMULOR_API_KEY` returns a non-empty value
* `python3 scripts/famulor_client.py list_assistants` returns API data
* No `401 Unauthorized` error appears

## Security notes

* Never commit API keys to Git.
* Prefer local environment variables or an ignored `.env` file.
* Rotate API keys immediately if exposed.

<Tip>
  Related pages: [Use Cases & Prompts](/en/mcp/use-cases), [MCP Server](/en/mcp/server), [MCP Client](/en/mcp/client), and [API integration examples](/en/developers/api-integration-examples).
</Tip>