OVOS-Persona

September 16, 2026 · View on GitHub

PersonaPipeline adds multi-persona management to OpenVoiceOS (OVOS). Personas are configurable virtual assistants: each one assigns its own set of solver plugins to answer queries, so you can customize how OVOS handles a conversation.

See docs/index.md for the architecture and API reference.

Quick Start

  1. Update the core and install the plugin: 
    pip install -U "ovos-core>=0.5.1" ovos-persona
  2. Install or update the plugins and skills a persona needs: 
    pip install -U skill-wolfie ovos-skill-wikipedia ovos-skill-wikihow skill-wordnet ovos-openai-plugin
  3. Uninstall the ChatGPT fallback skill, since ovos-persona replaces it: 
    pip uninstall skill-ovos-fallback-chatgpt
  4. Edit mycroft.conf. The "..." below is a placeholder for your existing pipeline entries, not literal text. 
    {
  "intents": {
      "persona": {
        "handle_fallback":  true,
        "default_persona": "Remote Llama"
      },
      "pipeline": [
          "stop_high",
          "converse",
          "ocp_high",
          "padatious_high",
          "adapt_high",
          "ovos-persona-pipeline-plugin-high",
          "ocp_medium",
          "...",
          "fallback_medium",
          "ovos-persona-pipeline-plugin-low",
          "fallback_low"
    ]
  }
}
  5. Restart OVOS.
  6. Check the logs to confirm the persona loaded without errors: 
    cat ~/.local/state/mycroft/skills.log | grep persona
  7. Read the Persona Intents section for the voice commands.

Features

  • Multiple personas: manage a list of personas, each with its own solver plugins.
  • Dynamic switching: activate a different persona at any time.
  • Per-session state: the active persona and conversation memory are tracked per session, so concurrent conversations stay isolated.
  • Short-term memory: a default short-term memory ships with ovos-persona and is always available. Swap it for any opm.agents.memory plugin through the memory_module config key. See docs/memory.md.
  • Conversational: personas can handle utterances directly, without a matching skill.
  • Personalize: create a persona with a simple .json file. See docs/defining-personas.md.

Installation

pip install ovos-persona

Persona Intents

The persona service supports voice intents for managing persona interactions: listing personas, checking the active persona, activating a persona, asking a persona a single question, and stopping the conversation. Each intent corresponds to a messagebus event.

See the OVOS technical manual: Persona Pipeline for the full list of example utterances and bus events.

Pipeline Configuration

Where you place "ovos-persona-pipeline-plugin-high" in the pipeline decides whether the active persona gets full control of an utterance, or only handles it after high-confidence skills fail to match. "ovos-persona-pipeline-plugin-low" handles utterances as a fallback even when no persona is explicitly active, replacing OpenVoiceOS/ovos-skill-fallback-chatgpt.

See the OVOS technical manual: Persona Pipeline for the pipeline configuration strategies and example mycroft.conf snippets.

Creating a Persona

Personas are configured with JSON files. A persona can come from:

  1. a plugin (for example, the OpenVoiceOS/ovos-openai-plugin), or
  2. a user-defined JSON file in ~/.config/ovos_persona.

Personas rely on solver plugins, which try to answer a query in sequence until one succeeds.

Example: a persona using a local OpenAI-compatible server. Save this as ~/.config/ovos_persona/llm.json:

{
  "name": "My Local LLM",
  "handlers": [
    "ovos-solver-openai-plugin"
  ],
  "ovos-solver-openai-plugin": {
    "api_url": "https://llama.smartgic.io/v1",
    "key": "sk-xxxx",
    "system_prompt": "helpful, creative, clever, and very friendly."
  }
}

A persona does not need an LLM. Simpler solvers work too, even without a GPU.

Example: OldSchoolBot, a persona built from non-LLM solvers.

{
  "name": "OldSchoolBot",
  "handlers": [
    "ovos-solver-wikipedia-plugin",
    "ovos-solver-ddg-plugin",
    "ovos-solver-plugin-wolfram-alpha",
    "ovos-solver-wordnet-plugin",
    "ovos-solver-rivescript-plugin",
    "ovos-solver-failure-plugin"
  ],
  "ovos-solver-plugin-wolfram-alpha": {"appid": "Y7353-xxxxxx"}
}

Behavior:

  • Searches online sources such as Wikipedia and Wolfram Alpha.
  • Falls back to offline word lookups through WordNet.
  • Uses a local chatbot (RiveScript) for chitchat.
  • The "failure" solver catches errors so the persona always returns a response.

HiveMind Integration

This project includes a native hivemind-plugin-manager integration for interoperability with the HiveMind ecosystem.

Contributing

Found a bug or have an idea? Open an issue or submit a pull request.

Credits

Developed by TigreGótico for OpenVoiceOS.

NGI0 Commons Fund

This project was funded through the NGI0 Commons Fund, a fund established by NLnet with financial support from the European Commission's Next Generation Internet programme, under the aegis of DG Communications Networks, Content and Technology under grant agreement No 101135429.