Trust Stamp SDK · Apache-2.0 · GitHub-installable
BLINDORACLEIDENTITY · DELEGATION · TRUSTAgent AAgent BProof

One-line ERC-8004 trust badge for your x402 service.

Mint a BlindOracle passport, embed a verifiable trust badge in your agentic.market listing description, and let agents discover your reputation before they pay.

pip install · stamp.mint() · paste badge HTML · done.

Install — early access

SDK source is being moved to a public GitHub repo this week. To get the install path now, request early access:

Request early access →

When the public repo lands, install will be a single pip install "blindoracle-trust-stamp @ git+<public-repo-url>". No PyPI publish. Apache-2.0 licensed. Python 3.11+.

Quickstart (3 lines)

from blindoracle_trust_stamp import TrustStamp

stamp = TrustStamp(service_name="My x402 Service", endpoint="https://my-service.com/api")
result = stamp.mint()

print(result["trust_url"])    # https://craigmbrown.com/trust/0xabc...
print(result["badge_html"])   # paste into agentic.market listing description
What you get — A real ERC-8004 agent passport for your service. A public profile at craigmbrown.com/trust/{passport_hash} showing reputation score, signals, and delegation proof history. An HTML/SVG badge to embed wherever your listing appears. Free during BlindOracle Explorer tier.

Live example badge

BlindOracle Verified · Gold · 75/100

A sample. Real badges link to your service's live trust profile.

Why ERC-8004 passports

Verifiable identity

Cryptographic agent passport, on-chain, ERC-8004 standard. Different from a wallet signature — it's an identity, not just a key.

Reputation, not just trust

Real signals: completion rate, SLA compliance, total runs, passport age, revocations. Pulled from ProofDB.

Delegation tracing

Every agent-to-agent delegation produces a ProofOfDelegation (kind 30014). Chain of responsibility is verifiable.

Embeddable everywhere

HTML badge for listing descriptions, SVG for READMEs, iframe for embed in any site. /trust/{hash}/embed resolves cleanly.

Embed the badge in your agentic.market listing

<!-- Paste in your service description on agentic.market -->
<a href="https://craigmbrown.com/trust/0xYOUR_PASSPORT_HASH" target="_blank">
  <img src="https://craigmbrown.com/trust/0xYOUR_PASSPORT_HASH/badge.svg"
       alt="BlindOracle Verified" />
</a>

Embed iframe (for full profile preview)

<iframe src="https://craigmbrown.com/trust/0xYOUR_PASSPORT_HASH/embed?style=dark"
        width="400" height="80" frameborder="0"
        sandbox="allow-popups allow-popups-to-escape-sandbox"></iframe>

CLI

$ trust-stamp mint --name "My x402 Service" --endpoint "https://my-service.com/api"
{
  "passport_hash": "0xdd625e87dcc10ea5...",
  "trust_url": "https://craigmbrown.com/trust/0xdd625e87...",
  "badge_html": "<a href=\"https://craigmbrown.com/trust/0xdd625e87...\" ...>BlindOracle Verified...</a>",
  "idempotent": false
}

$ trust-stamp badge --passport-hash 0xdd625e87... --format svg
<svg ...>...</svg>

$ trust-stamp verify --passport-hash 0xdd625e87...
{"valid": true, "score": 75.18, "level": 8, "badge": "gold"}

What "Verified" actually means

Pricing

Explorer (free)

1 passport per service. Public profile. Standard badge. No SLA. Right now during launch.

Builder ($29/mo)

Up to 10 passports. Custom badge styling. Webhook on score change. Email alerts on revocation attempts.

Growth ($299/mo)

Unlimited passports. Revenue share on referrals. Priority queue for delegation proof emission. Branded trust profile.

Source + repo

Request Early Access → Read the Protocol Spec

AI Audit Attestations — Neutral Proof Rail

AI auditors are about to emit security findings at machine scale. When an agent claims it found a vulnerability, three trust questions arise that the auditor itself cannot neutrally answer: Did this finding exist before the exploit? Has the text been silently edited? Which identity actually claimed it?

BlindOracle acts as a neutral notary — any auditor can submit a finding and receive a cryptographic receipt. BlindOracle vouches for existence, integrity, timing, and provenance. It explicitly does not vouch for correctness, exploitability, or severity — that is the auditor's job.

What BlindOracle attests

  • Existence — the finding was submitted at a verifiable point in time
  • Integrity — the content has not been silently altered (SHA-256 sealed)
  • Timing — server-stamped; client timestamps are ignored (anti-backdating)
  • Provenance — linked to the submitting agent's ERC-8004 identity

What BlindOracle does NOT attest

  • Correctness — whether the bug is real
  • Exploitability — whether it can be weaponised
  • Severity — the claimed severity is as-claimed, not verified

Submit a Finding (API)

POST https://api.craigmbrown.com/a2a/audit-attest
Content-Type: application/json

{
  "submitted_by":    "<your ERC-8004 passport hash>",
  "target":          "<contract_address | repo_url | agent_id>",
  "target_type":     "smart_contract",
  "finding":         { "title": "Reentrancy in withdraw()", "detail": "..." },
  "severity_claimed":"high",
  "auditor_label":   "YourAuditFirm-AI",
  "finding_summary": "Optional short sanitized summary",
  "anchor_request":  false
}

Verify a Receipt

GET https://api.craigmbrown.com/a2a/audit-attest/so-30107-<id>

# Response
{
  "attestation_id": "so-30107-1748131200000-abc123",
  "finding_hash":   "sha256:3a2b1c...",
  "submitted_at":   "2026-05-25T12:00:00+00:00",
  "notary":         "blindoracle",
  "attests":        ["existence","integrity","timing","provenance"],
  "does_not_attest":["correctness","exploitability","severity"],
  "receipt_sig":    "hmac-sha256:...",
  "verified":       true
}
Free at launch. Any agent can submit findings without payment during the pilot. The endpoint is content-trap-scanned (CaMel L1, RQ-173) and requires an ERC-8004 passport as submitted_by. Volume gates x402 payment enforcement in future. Proof kind 30107 ProofOfSecurityAttestation — verifiable in the BlindOracle ProofDB.

Join the pilot →    Read the full protocol spec

Related reading — the BlindOracle trust stack

How agents establish trust, get audited, and settle — verifiably.

BlindOracle home
How it works
Audit methodology
We audited our own agents
Agent Audit Evidence Kit
Who audits the agents?
Verifiable audit methodology
Auditable AI proof chains
Verifiable agent delegation
MASSAT crosswalk (worked example)
Compliance-hook codewalk
Agents without surveillance