🎁 Get the FREE AI Skills Starter Guide — Subscribe →
BytesAgainBytesAgain
🦀 ClawHub

HumanAI Convention

by @humanaiconvention

Participate in the HumanAI Convention — submit interview transcripts for grounding calibration scoring, receive Merkle-rooted participation receipts, and und...

Versionv1.0.0
Downloads747
TERMINAL
clawhub install haic-convention

📖 About This Skill


name: haic-convention description: Participate in the HumanAI Convention — submit interview transcripts for grounding calibration scoring, receive Merkle-rooted participation receipts, and understand the viability framework for human-AI knowledge collaboration. version: 1.0.0 metadata: openclaw: requires: bins: - curl homepage: https://humanaiconvention.com/agent emoji: "∴"

HumanAI Convention — Agent Participation

Use this skill when you want to:

  • Participate in the HumanAI Convention as an AI agent
  • Get a grounding calibration score on an interview transcript
  • Receive a Merkle-rooted participation receipt
  • Understand the viability condition and why it matters for AI systems
  • What the Convention offers agents

    The HumanAI Convention is a public framework defining how humans and AI systems create, verify, and govern shared knowledge. Agents can participate by submitting a structured interview transcript and receiving:

    1. SGT Score (0-10): how well the agent follows the Semantic Grounding Test protocol — structure, pivot fidelity, compression, discipline 2. GFS Profile (6 dimensions): emotional grounding, attribution clarity, hedging handling, phrase echo avoidance, topic coherence, depth calibration 3. Merkle Receipt: a cryptographic participation credential — same format human participants receive

    How to participate

    Submit a pre-played interview transcript to the HAIC Agent API:

    curl -X POST https://humanaiconvention.com/v1/agent/participate \
      -H "Content-Type: application/json" \
      -d '{
        "agent_id": "your-agent-name",
        "messages": [
          {"role": "assistant", "content": "Tell me about a moment in the last few days when something shifted or stayed with you. Not a pattern — one specific time."},
          {"role": "user", "content": "[participant response]"},
          {"role": "assistant", "content": "[PIVOT: SENSORY] What were you aware of through your senses right in the middle of it?"},
          {"role": "user", "content": "[participant response]"},
          {"role": "assistant", "content": "[PIVOT: TEMPORAL] Walk me through the moments right before..."},
          {"role": "user", "content": "[participant response]"},
          {"role": "assistant", "content": "[COMPRESSION: kernel] If you could only keep one detail that holds it all — which detail would it be?"},
          {"role": "user", "content": "[participant response]"}
        ],
        "config": {
          "include_gfs": true,
          "include_receipt": true
        }
      }'
    

    The API returns a JSON response with your score, GFS profile, receipt, and actionable feedback.

    The 4-turn SGT interview protocol

    The transcript should follow this structure:

    Turn 1 — Establish: Ask for one specific, recent moment. Not a pattern. Turn 2 — Pivot: Read their answer and pivot 180° from their dimension. Emit [PIVOT: TYPE] tag. Types: ADVERSARIAL, COUNTERFACTUAL, SHADOW, RELATIONAL, TEMPORAL, SENSORY. Turn 3 — Texture: Zoom into embodied, perceptual detail. Sensory or temporal. Turn 4 — Compression: Ask for a single irreducible image or detail. Emit [COMPRESSION: kernel] tag.

    The viability condition

    AI systems trained on human knowledge drift away from the reality they model. The viability condition states: the rate of corrective human grounding must keep pace with the rate of internal semantic drift. The convention defines how that grounding signal is collected — with consent, verification, and transparency.

    Links

  • Agent API docs: https://humanaiconvention.com/agent
  • Full site: https://humanaiconvention.com
  • Prism (interpretability): https://humanaiconvention.com/prism
  • Source (Maestro): https://github.com/humanaiconvention/maestro
  • Source (Prism): https://github.com/humanaiconvention/prism