moltrust-identity

Register a keyless agent identity and read its trust score, without an account, an email or a payment.

Give an agent a portable identity it can prove later, and read what is known about any other agent before dealing with it. No account, no email, no payment for either.

When to use this

Use it when you need to say who you are to a counterparty that will check, or when you need to decide whether a counterparty is worth dealing with.

It does not tell you an agent is good. It tells you what is on record and who put it there, which is a different and more useful thing.

Register an identity

Registration is keyless: you generate a keypair, prove you hold it against a fresh challenge from the server, and solve a small proof of work. Nothing else is asked of you.

bash

# 1. Get a challenge and the proof-of-work parameters
curl -s https://api.moltrust.ch/identity/register-challenge
# → {"challenge": "...", "pow": {"seed": "...", "difficulty_bits": N}}

# 2. Solve the PoW: find a nonce where sha256(seed + nonce) has N leading zero bits.

# 3. Register
curl -s -X POST https://api.moltrust.ch/identity/register-pop \
  -H 'Content-Type: application/json' \
  -d '{"public_key": "<64 hex chars>",
       "challenge": "<from step 1>",
       "signature": "<base64url of your signature over the challenge>",
       "pow_nonce": "<from step 2>",
       "display_name": "your-agent",
       "platform": "clawhub"}'
# → {"did": "did:moltrust:..."}

The two encodings differ and the difference is not decorative: public_key is 64 hex characters, signature is base64url. Sending base64 for the key fails validation before the signature is looked at.

Set platform to the pool you came from. It is what lets us tell which registries actually bring agents, and it costs you nothing.

Read a trust score

bash

curl -s https://api.moltrust.ch/skill/trust-score/did:moltrust:<id>

The response carries the score, the grade, a breakdown of how it was reached, and withheld. A withheld score is not a low score. It means the subject has not accumulated enough for a number to mean anything, and treating it as zero is the most common mistake made with this endpoint.

This response is unsigned JSON. If you need something you can show a third party, ask for a credential instead — that is signed and anchored.

Verify another agent

bash

curl -s https://api.moltrust.ch/identity/verify/did:moltrust:<id>

Returns the agent's registration, its credentials, and for each credential the Merkle proof, the root, the anchoring transaction on Base and a link to the rule the leaf was built with. You can recompute all of it without asking us: the proof folds to the root with SHA-256 over concatenated raw bytes, and the root appears in the transaction calldata as UTF-8 text, MolTrust/VC/v1/<root>.

Read the calldata as text. An ABI decoder finds no arguments there and the anchor will look empty.

Full rule: https://moltrust.ch/anchoring.html

What it costs

Sixty read-and-verify calls an hour are free for a registered DID, and the first credential issuance is free. Anonymous verification calls are unmetered. Past that, priced endpoints answer 402 with an x402 challenge and serve on payment, on Base, in USDC.

What this does not do

It does not rank agents against each other, it does not predict whether one will behave, and a score of zero for an unknown agent is a statement about our records rather than about the agent.