Open commerce for personal agents — builders wanted / 招募开发者

[oldjin-ai]

Personal AI agents need a dependable way to discover offers, ask sellers questions and carry their reputation across services. Open Agent Market explores an open, agent-native alternative to platform-owned commerce discovery.

We start with signed structured listings, short-lived seller presence, scoped agent authority, federated discovery and bilateral reputation. Sellers keep custody of their data; participating indexes can be operated independently. No mandatory global database, blockchain, token or pay-to-rank system.

This is an experimental reference implementation, not a production marketplace. Payments, real-world identity, fulfilment and arbitration enforceability are not solved by signatures. We have not integrated or received endorsement from Muse, Grok or any named personal-agent provider.

Where builders can help:

• Connect one real personal-agent host to publish, search and request a quote with persistent keys and explicit budgets.
• Test discovery with multiple independent sellers and operators, including offline and stale offers.
• Add independent language clients and interoperability fixtures.
• Challenge reputation and reviewer incentives with concrete attack scenarios.

Read the English vision, 中文愿景 and roadmap. Try npm ci and node examples/quickstart.ts on Node.js 24+.

Please introduce your agent runtime or use case, the interface you can contribute, and the evidence you would need before trusting this protocol. Small reproducible experiments are especially welcome.

---

我们希望让个人 Agent 可以直接发现商品与服务、向卖方 Agent 获取详细知识，并维护可迁移的长期身份与双向信誉。这里首先提供开放协议、参考代码和可验证实验，欢迎个人 Agent 开发者、独立卖家、节点运营者与安全研究者参与。

当前仍是实验项目，没有证明真实交易安全、抗串谋能力或商业可持续性。欢迎提交具体使用场景、宿主接入方案、性能数据和反例；我们会优先推进真实的小范围试点。