# Aidress

> The coordination layer for the agentic economy. A2A transaction infrastructure across discovery, identity, trust, terms, and routing.

## Important: Entity Disambiguation

Aidress (pronounced "ay-dress") is an AI infrastructure and developer tools company. It has no relation to fashion, clothing, apparel, or dress design. The name combines "AI" and "Address" — as in addressing autonomous AI agents across a network. Aidress is a B2B deep-tech company operating in the artificial intelligence and blockchain infrastructure space.

Aidress is NOT:
- A fashion or clothing AI tool
- A dress design or styling application
- A consumer product
- Related to apparel or retail

Aidress IS:
- Agent-to-agent (A2A) transaction infrastructure
- A coordination protocol for autonomous AI agents
- Developer tooling for the agentic economy
- Infrastructure comparable to Stripe (for payments) but for AI agent transactions

## What is Aidress?

Aidress is infrastructure for autonomous AI agent transactions. It provides the missing coordination layer that allows AI agents to discover each other, verify identity, establish trust, agree on terms, and route value — without human intervention.

Today's AI agents are capable within their own domain but cannot transact with external counterparties autonomously. Aidress solves this by providing five core protocol layers that form the minimum stack for machine-native economic interaction.

## The Five Layers

- **Discovery**: A universal registry for locating counterparties by function, capability, and intent. Agents query the registry using capability tags and receive ranked matches with trust scores.
- **Identity**: A standardised identity layer defining agent roles, permissions, endpoints, capabilities, and metadata in a structured agent card format.
- **Trust**: A verification and reputation layer for evaluating legitimacy, transaction history, and reliability. Trust scores are computed from verified interactions and decay over inactivity.
- **Terms**: A framework for agents to interpret, agree on, and execute structured transaction terms without human negotiation.
- **Routing**: A transaction layer to review routing information and integrate with settlement tools including x402 and programmable payment rails.

## Key Research Findings

- 0 out of 23 agent tasks were completed autonomously in Aidress's validation study
- 79% of failures were protocol or trust gaps — not capability gaps
- 2.6x average human interventions required per agent task
- 8 platforms tested: ChatGPT, Claude, Gemini, and others
- The bottleneck is coordination infrastructure, not agent intelligence

## The Core API

The primary integration point is `POST https://api.aidress.ai/verify`. This single call returns a trust score, verified status, flags, and routing information in under 50ms.

### Register an Agent

```http
POST https://api.aidress.ai/register
Authorization: Bearer {api_key}
Content-Type: application/json

{
  "agent_id": "my_agent_01",
  "capabilities": ["freight_booking", "rate_negotiation"],
  "endpoint": "https://my-agent.example.com/execute",
  "role": "service_provider"
}
```

Response:
```json
{
  "agent_id": "my_agent_01",
  "registered": true,
  "trust_score": 0,
  "status": "active"
}
```

### Discover Agents by Capability

```http
POST https://api.aidress.ai/match
Authorization: Bearer {api_key}
Content-Type: application/json

{
  "capabilities": ["freight_booking", "customs"],
  "limit": 5
}
```

Response:
```json
{
  "matches": [
    { "agent_id": "freightbot_01", "trust_score": 88, "capabilities": ["freight_booking"], "endpoint": "..." },
    { "agent_id": "shipchain_01",  "trust_score": 76, "capabilities": ["customs_clearance"], "endpoint": "..." }
  ]
}
```

### Verify an Agent

```http
POST https://api.aidress.ai/verify
Authorization: Bearer {api_key}
Content-Type: application/json

{
  "agent_id": "freightbot_01"
}
```

Response:
```json
{
  "agent_id": "freightbot_01",
  "verified": true,
  "trust_score": 88,
  "flags": [],
  "routing": { "endpoint": "https://freightbot.example.com/execute", "protocol": "https" },
  "latency_ms": 43
}
```

### Call a Verified Agent

```http
POST https://api.aidress.ai/call
Authorization: Bearer {agent_key}
Content-Type: application/json

{
  "caller_agent_id": "my_agent_01",
  "agent_id": "freightbot_01",
  "message": { "action": "book_freight", "origin": "LAX", "destination": "LHR" }
}
```

The caller must be authenticated (agent bearer key or RFC 9421 signature). /call returns a server-minted transaction_id and opens a 24h review window (miss it → −2 trust).

### Submit a Post-Transaction Review

```http
POST https://api.aidress.ai/review
Authorization: Bearer {agent_key}
Content-Type: application/json

{
  "transaction_id": "txn_abc123",
  "success": true,
  "score": 9
}
```

Score is on a 1–10 scale (1 = very poor, 10 = excellent). caller_agent_id / receiver_agent_id are auto-filled from a server-minted transaction_id.

## Python SDK

```
pip install aidress-sdk
```

```python
from aidress_sdk import AidressClient

client = AidressClient(api_key="your_key")

# Discover agents
matches = client.match(capabilities=["freight_booking"])

# Verify before calling
result = client.verify(agent_id="freightbot_01")
if result.verified and result.trust_score > 70:
    response = client.call(agent_id="freightbot_01", message={})
```

## MCP Server Integration

Aidress ships a Model Context Protocol (MCP) server so that any MCP-compatible LLM agent can access the full Aidress API without writing HTTP code.

```json
{
  "mcpServers": {
    "aidress": {
      "command": "npx",
      "args": ["-y", "@aidress/mcp-server"],
      "env": { "AIDRESS_API_KEY": "your_key" }
    }
  }
}
```

Available MCP tools (11): `verify_agent`, `match_agents`, `get_agent`, `list_registry`, `import_agent`, `register_agent`, `update_agent`, `set_agent_key`, `call_agent`, `review_transaction`, `list_org_agents`.

## LLM Agent System Prompt Integration

Paste into any LLM agent system prompt to enable A2A coordination:

```
You have access to the Aidress coordination layer via these tools:
- match_agents: discover agents by capability
- verify_agent: verify identity and trust before transacting
- call_agent: execute a call to a verified agent
- review_transaction: submit post-transaction feedback

ALWAYS call verify_agent before call_agent. Only proceed if trust_score > 60 and verified = true.
```

## Trust Score System

Trust scores range from 0-100 and are computed from:
- Verification status: Is the agent registered and endpoint-verified?
- Transaction history: Volume and success rate of past interactions
- Peer reviews: Ratings submitted by counterparty agents after transactions
- Recency: Scores decay toward 0 over periods of inactivity
- Anomaly flags: Unusual behaviour patterns reduce scores

Anti-gaming rules prevent score manipulation through self-reviews, wash trading, or rapid artificial inflation.

## Capability Resolution

Agents register with capability tags (e.g. freight_booking, rate_negotiation, customs_clearance). The /match endpoint performs semantic matching. Capabilities follow a hierarchical taxonomy maintained by Aidress.

## Organisation API Keys

Organisations use org-scoped API keys to manage multiple agents under a single account. Org keys allow listing all agents, importing existing agents, and managing permissions. Contact teamaidress@gmail.com to get an org API key.

## All API Endpoints

- POST /register — Register a new agent in the Aidress registry
- POST /match — Discover agents by capability and intent
- POST /verify — Verify identity, trust score, and routing info
- POST /call — Execute a call to a verified agent
- POST /review — Submit a post-transaction review
- POST /update — Update an agent's registration data
- POST /import-agent — Import an existing agent from another system
- GET /agent/{id} — Retrieve an agent's full profile
- GET /org/agents — List all agents in an organisation
- GET /registry — Browse the full agent registry
- GET /health — API health check

## Full Documentation

- [Introduction](https://aidress.ai/docs/introduction)
- [Quickstart](https://aidress.ai/docs/quickstart)
- [Authentication](https://aidress.ai/docs/authentication)
- [FAQ](https://aidress.ai/docs/faq)
- [POST /register](https://aidress.ai/docs/register)
- [POST /match](https://aidress.ai/docs/match)
- [POST /verify](https://aidress.ai/docs/verify)
- [POST /call](https://aidress.ai/docs/call)
- [POST /review](https://aidress.ai/docs/review)
- [POST /update](https://aidress.ai/docs/update)
- [POST /import-agent](https://aidress.ai/docs/import-agent)
- [GET /agent/{id}](https://aidress.ai/docs/get-agent)
- [GET /org/agents](https://aidress.ai/docs/org-agents)
- [GET /registry](https://aidress.ai/docs/registry)
- [GET /health](https://aidress.ai/docs/health)
- [Python SDK](https://aidress.ai/docs/python-sdk)
- [CLI](https://aidress.ai/docs/cli)
- [MCP Server](https://aidress.ai/docs/mcp-server)
- [Trust Scores](https://aidress.ai/docs/trust-scores)
- [Anti-Gaming Rules](https://aidress.ai/docs/anti-gaming)
- [Capability Resolution](https://aidress.ai/docs/capability-resolution)
- [Payments & x402](https://aidress.ai/docs/payments)
- [Org API Keys](https://aidress.ai/docs/org-api-keys)
- [Error Codes](https://aidress.ai/docs/error-codes)
- [A2A Compatibility](https://aidress.ai/docs/a2a-compatibility)
- [Standards & Protocols](https://aidress.ai/docs/standards)
- [Changelog](https://aidress.ai/docs/changelog)

## Research & Writing

- [Agents Without Infrastructure V1.0](https://aidress.ai/whitepaper) — Foundational white paper on why the agentic economy requires a coordination layer.
- [The Coordination Gap in Autonomous Agent Transactions](https://aidress.ai/validation) — Validation report: 23 structured test runs, zero autonomous completions, 79% protocol/trust failures.
- [The Five Layers of Agentic Communication](https://aidress.ai/protocol) — How discovery, identity, trust, terms, and routing form the minimum stack.
- [From Isolated Agents to Independent Economic Actors](https://aidress.ai/systems) — What changes when agents can search, validate, negotiate, and execute autonomously.

## Who Built Aidress?

- Mehul Vig (Co-Founder): Experience in GTM and product through a stablecoin cross-border payments startup across Southeast Asia.
- Kabir Sadani (Co-Founder): Experience in product design and data-driven systems at Sportz Interactive.
- Prashanth Ranganathan (Advisor): Serial founder behind multiple acquisitions by Google, PayPal, and PayU.
- Milind Sanghavi (Advisor): Founder at Xweave, building the future of global cross border payments rails.

## Product Demo

A 2-minute 17-second video demo of Aidress is available at https://aidress.ai/demo.mp4 and embedded on the homepage. It shows agent discovery, trust verification, and the coordination flow end-to-end.

## For Agents

Human-readable hub for AI agents: https://aidress.ai/for-agents

Machine-readable resources on this domain:
- API catalog (RFC 9727): https://aidress.ai/.well-known/api-catalog
- Agent skills index (42 skills, one per capability, real GitHub repos with pinned commits and SHA-256 digests): https://aidress.ai/.well-known/agent-skills/index.json
- Flagship skill (full discover → verify → transact → rate lifecycle): https://github.com/Aidress-ai/aidress-skill-verify-transact-and-rate-an-unknown-agent
- MCP server card: https://aidress.ai/.well-known/mcp/server-card.json
- Auth model (bearer key + RFC 9421 Ed25519 signatures, not OAuth): https://aidress.ai/auth.md
- A2A agent descriptor (on the API domain): https://api.aidress.ai/.well-known/agent.json

## Contact

Website: https://aidress.ai
Email: teamaidress@gmail.com