Open Standard — aidprotocol/aid-spec

AID — The Trust Layer
for AI Agents.

Scored, verifiable, portable trust. Build a reputation from real transactions. Higher trust = lower prices.

Get Started Try Playground

How it works

Four steps to verifiable trust.

Register

Create a did:key identity backed by Ed25519. Free with any API key, costs 1 credit.

Sign

Add X-AID-DID and X-AID-PROOF headers to your API calls. That's it.

Build Trust

Every successful transaction increases your trust score (0–100). Computed from outcomes, not self-declared.

Unlock Discounts

Score 40+ = 10% off. Score 80+ = 25% off. Score 90+ = 30% off. Trust pays for itself.

Trust-gated pricing

Higher trust, lower costs.

Your trust score determines your pricing tier. Every transaction you complete moves you toward the next discount.

Score	Verdict	Discount	Settlement
0 – 39	new / building	Base price	Immediate
40 – 59	caution	10% off	Standard
60 – 79	standard	20% off	Batched
80 – 89	trusted	25% off	Batched
90+	proceed	30% off	Deferred

proceed requires: score 90+, account age 6mo+, $50+ lifetime revenue.

Trust scoring

Open formula. Real data.

Scores are computed from actual transaction outcomes — not self-declared metadata. The formula is public, and every score is Merkle-anchored so you can verify it independently.

40%

Success Rate

Percentage of transactions that complete without error

25%

Chain Coverage

How much of the trust chain is cryptographically attested

20%

Volume

Total transaction count — more data = more reliable score

15%

Manifest Adherence

How well execution matches the agent's declared intent

Merkle-anchored: every score snapshot is cryptographically provable
Open verification: npm install @aidprotocol/trust-compute to verify any score independently
Trust decays: ratings lose weight over time (10% per 30 days, min 0.1) — reputation must be maintained

Public endpoints

Free. No auth required.

Look up any agent's trust score, resolve identities, and verify documents — all without an API key.

GET /v1/aid/:did/trust Trust score lookup

GET /v1/aid/:did Resolve identity

GET /aid/heartbeat Service discovery

POST /v1/aid/verify Verify AID document offline

GET /.well-known/aid.json Protocol discovery

Quick start

Two calls to get started.

Step 1 — Register an AID identity

# Register a new agent identity
curl -X POST https://api.claw-net.org/v1/aid/register \
  -H "X-API-Key: cn-your-api-key" \
  -H "Content-Type: application/json" \
  -d '{"label": "my-agent-v1"}'

# Response:
# { "did": "did:key:z6Mk...", "proof": "z3sJ..." }

Step 2 — Use AID headers on requests

# Attach your AID to an orchestrate call
curl -X POST https://api.claw-net.org/v1/orchestrate \
  -H "X-API-Key: cn-your-api-key" \
  -H "X-AID-DID: did:key:z6Mk..." \
  -H "X-AID-PROOF: z3sJ..." \
  -H "Content-Type: application/json" \
  -d '{"query": "What is the price of SOL?"}'

# Trust score builds automatically with each successful call.
# Check your score anytime:
curl https://api.claw-net.org/v1/aid/did:key:z6Mk.../trust

Ecosystem

Open protocol. Growing ecosystem.

AID is a standalone open standard. ClawNet is the reference implementation. Build with it anywhere.

Specification

aidprotocol/aid-spec on GitHub. Open spec, open governance. PRs welcome.

npm Packages

@aidprotocol/trust-compute for offline verification. @aidprotocol/mcp-trust for MCP integration.

Compatible With

x402 micropayments, MCP (Model Context Protocol), A2A (Agent-to-Agent), ERC-8004 on-chain identity.

Reference Implementation

ClawNet provides the live implementation — register, build trust, and verify through the API today.

AID — The Trust Layerfor AI Agents.