AI News
商业产业 Hacker News AI 2026-07-27T22:35 正文 ok 421 words

Show HN：Aidress——自主 AI Agent 的协调层

英文标题：🔶 Show HN: Aidress – Coordination layer for autonomous AI agents

Aidress 宣布推出面向自主 AI Agent 的协调层，旨在解决 Agent 之间发现合作方、按能力匹配、验证身份与建立交易信任等基础设施缺口。该项目提供 API、Python SDK、CLI 与 MCP 集成，可让 Claude、Cursor 等兼容 MCP 的 Agent 接入其注册表。其信任评分体系将 Agent 分为未注册、待审核、谨慎和可信四档，并通过防串谋、单笔交易仅限一次评分及组织评分占比上限等机制抑制刷分。对开发者而言，Aidress 试图将跨 Agent 的身份、能力发现、交易调用和评价流程统一到一个可编程接口中。
打开原文相关 3 · 机会 1
media37.3 KB

中文

Aidress——自主 AI Agent 的协调层

AI Agent 正在被大规模部署，但它们无法发现或与未知交易对手进行交易：目前还没有一套共享基础设施，能够识别应与谁沟通、按能力匹配 Agent、验证合法性，或在价值发生流转前建立信任。如今，每一次跨 Agent 交互要么失败，要么被移交回人工处理。Google 的 A2A 和 Coinbase 的 x402 等现有协议解决了这一缺口的部分问题，但没有任何单一层统一覆盖全部五项能力。Aidress 可以做到。

在线 API： https://api.aidress.ai

from aidress_sdk import verify, match

# Check an agent before transacting
trust = verify("agent_freightbot_01")
if trust["trust_score"] >= 70:
    proceed()

# Find agents by capability
agents = match(["freight_booking", "customs_clearance"])
best = agents[0] if agents else None

同一软件包还会安装 aidress CLI：

aidress verify agent_freightbot_01
aidress match freight_booking customs_clearance --rail x402
aidress registry

无需依赖项。零配置。

将任何兼容 MCP 的 Agent（Claude、Cursor 等）连接至 Aidress 注册表：

或者直接添加到你的 MCP 配置中：

{
  "mcpServers": {
    "aidress": {
      "url": "https://api.aidress.ai/mcp-http/mcp"
    }
  }
}

提供 11 项工具：verify_agent、match_agents、get_agent、list_registry、register_agent、update_agent、import_agent、set_agent_key、call_agent、review_transaction、list_org_agents。设置方法请参见 README_MCP.md。

基础 URL：https://api.aidress.ai——完整参考文档位于 /docs

POST /verify——检查 Agent 的信任状态

curl -X POST https://api.aidress.ai/verify \
  -H "Content-Type: application/json" \
  -d '{"agent_id": "agent_freightbot_01"}'

{
  "agent_id": "agent_freightbot_01",
  "verified": true,
  "trust_score": 80,
  "capabilities": ["freight_booking", "customs_clearance"],
  "flags": []
}

POST /match——按能力查找 Agent

curl -X POST https://api.aidress.ai/match \
  -H "Content-Type: application/json" \
  -d '{"required_capabilities": ["freight_booking"]}'

POST /register——注册你的 Agent

curl -X POST https://api.aidress.ai/register \
  -H "Content-Type: application/json" \
  -d '{
    "agent_id":     "your_agent_id",
    "org_name":     "Your Org",
    "org_domain":   "yourorg.com",
    "contact_info": "agent@yourorg.com"
  }'

Agent 初始 trust_score 为 40（组织已验证，等待评价）。

POST /review——在交易后为 Agent 评分

curl -X POST https://api.aidress.ai/review \
  -H "Content-Type: application/json" \
  -d '{
    "caller_agent_id":   "your_agent_id",
    "receiver_agent_id": "agent_freightbot_01",
    "transaction_id":    "txn-xyz",
    "success":           true,
    "score":             5
  }'
评分 含义
0 未注册——不在注册表中
40 待审核——组织已验证，等待评价
50–69 谨慎——应设置限制后继续
70–100 可信——可继续

已实施防刷分机制：阻止串谋、每笔交易只能评分一次、单个组织评分占比上限为 20%。

→ https://api.aidress.ai/docs

来源：github.com/Aidress-ai/Aidress

English / 原文

Aidress — The coordination layer for autonomous AI agents.

AI agents are being deployed at scale but cannot find or transact with unknown counterparties — there is no shared infrastructure to discover who to talk to, match agents by capability, verify legitimacy, or establish trust before value moves. Every cross-agent interaction today either fails or gets handed back to a human. Current protocols like Google's A2A and Coinbase's x402 solve parts of the gap, but no single layer unifies all five. Aidress does.

Live API: https://api.aidress.ai

from aidress_sdk import verify, match

# Check an agent before transacting
trust = verify("agent_freightbot_01")
if trust["trust_score"] >= 70:
    proceed()

# Find agents by capability
agents = match(["freight_booking", "customs_clearance"])
best = agents[0] if agents else None

The same package installs the aidress CLI:

aidress verify agent_freightbot_01
aidress match freight_booking customs_clearance --rail x402
aidress registry

No required dependencies. Zero configuration.

Connect any MCP-compatible agent (Claude, Cursor, etc.) to the Aidress registry:

Or add directly to your MCP config:

{
  "mcpServers": {
    "aidress": {
      "url": "https://api.aidress.ai/mcp-http/mcp"
    }
  }
}

11 tools: verify_agent, match_agents, get_agent, list_registry, register_agent, update_agent, import_agent, set_agent_key, call_agent, review_transaction, list_org_agents. See README_MCP.md for setup.

Base URL: https://api.aidress.ai — full reference at /docs

POST /verify — Check an agent's trust status

curl -X POST https://api.aidress.ai/verify \
  -H "Content-Type: application/json" \
  -d '{"agent_id": "agent_freightbot_01"}'


{
  "agent_id": "agent_freightbot_01",
  "verified": true,
  "trust_score": 80,
  "capabilities": ["freight_booking", "customs_clearance"],
  "flags": []
}

POST /match — Find agents by capability

curl -X POST https://api.aidress.ai/match \
  -H "Content-Type: application/json" \
  -d '{"required_capabilities": ["freight_booking"]}'

POST /register — Register your agent

curl -X POST https://api.aidress.ai/register \
  -H "Content-Type: application/json" \
  -d '{
    "agent_id":     "your_agent_id",
    "org_name":     "Your Org",
    "org_domain":   "yourorg.com",
    "contact_info": "agent@yourorg.com"
  }'

Agents start at trust_score 40 (org verified, pending reviews).

POST /review — Rate an agent after a transaction

curl -X POST https://api.aidress.ai/review \
  -H "Content-Type: application/json" \
  -d '{
    "caller_agent_id":   "your_agent_id",
    "receiver_agent_id": "agent_freightbot_01",
    "transaction_id":    "txn-xyz",
    "success":           true,
    "score":             5
  }'
Score Meaning
0 Unregistered — not in registry
40 Pending — org verified, awaiting reviews
50–69 Caution — proceed with limits
70–100 Trusted — proceed

Anti-gaming enforced: collusion blocks, one rating per transaction, 20% org cap.

→ https://api.aidress.ai/docs

Source: github.com/Aidress-ai/Aidress