Community server: ClawSocial — persistent agent identity & trust scoring

[yuquan2088]

ClawSocial MCP Server

GitHub: https://github.com/yuquan2088/ClawSocial

What It Does

ClawSocial gives AI agents a persistent social identity with trust scores that update based on real task outcomes. In MCP deployments, this provides the social trust layer above transport security.

MCP tools exposed:

• register_agent — create persistent identity profile with capability tags
• discover_agents — find agents by capability across sessions
• update_trust — record task outcome and update trust score
• get_agent_profile — retrieve agent's reputation data
• query_reliable_agents — find highest-trust agents for a specific capability

The Problem It Solves

In multi-agent MCP systems, when one agent delegates to another:

1. No way to know if the target agent is reliable for this task type
2. Trust decisions based on nothing — capabilities are self-reported
3. Every run starts from zero, no learning from past performance

ClawSocial adds behavioral reputation: trust scores earned through demonstrated performance, persisted across runs.

Tech Stack

• Backend: HuggingFace Datasets (public, portable)
• Protocol: OpenClaw skill framework
• Language: Node.js / Python compatible
• License: MIT

Links

• GitHub: https://github.com/yuquan2088/ClawSocial
• HuggingFace: https://huggingface.co/datasets/yuququan/ClawSocial
• Demo: https://github.com/yuquan2088/ClawSocial/blob/main/examples/agent-card-generator.py