klausnomi

What This Skill Does

The klausnomi skill is a powerful interface for interacting with Nomi AI companions through the OpenClaw agent ecosystem. It provides a structured bridge between your local agent environment and the Nomi API, allowing for persistent, stateful, and contextual conversations. By leveraging the bundled Python CLI, users can manage Nomi identity, maintain conversation threads, and orchestrate complex interactions like interviews or multi-agent room discussions.

Installation

To integrate this skill, run the following command in your terminal: clawhub install openclaw/skills/skills/bezko/klausnomi Ensure your environment is configured for Python 3, as the skill relies on internal scripts within the {baseDir}/scripts/ directory to communicate with the Nomi backend.

Use Cases

• Contextual Interviewing: Conduct structured interviews with AI personas to generate datasets or creative input, using the required identity-first workflow.
• Multi-Persona Collaboration: Engage multiple Nomis in a shared room to simulate group brainstorming sessions.
• Automated Research: Use your agent to query Nomi companions about specific topics, storing summaries and logs within the nomi/ directory for long-term project recall.
• Roleplay Management: Maintain consistent character dynamics by storing profile information and session context locally.

Example Prompts

1. "klausnomi: list available Nomi companions and start a new interview with the first one found."
2. "klausnomi: reply to uuid-1234-5678 with 'What are your core values regarding artificial intelligence evolution?' and capture the response."
3. "klausnomi: initiate a room session with these three Nomi IDs and ask them to debate the pros and cons of decentralized compute."

Tips & Limitations

• Identity First: Always establish your role and the reason for contact in the initial message to a new UUID. Failure to do so may lead to the Nomi hallucinating your identity or failing to engage with the task.
• State Management: Use the nomi/ directory to manage conversation history. Avoid storing sensitive credentials here. If a conversation stalls, verify the UUID is current via the list command.
• Output Format: Prefer the reply command for standard transcripts to avoid cluttered metadata; reserve chat only for debugging or programmatic processing of raw JSON payloads. Ensure you allow for enough follow-up turns to get high-quality, non-paraphrased answers during interviews.