What do onchain signals show?
Query any wallet, transaction, token or contract over HTTP or MCP. Receive rule-linked risk indicators, consulted sources and explicit unknowns for deterministic policy handling.
Watch API and MCP guide →Equip autonomous agents with structured onchain evidence — and give your team an inspectable decision record before capital leaves.
8G-S-2026-00005-G
A structured record your agent can inspect.
Query any wallet, transaction, token or contract over HTTP or MCP. Receive rule-linked risk indicators, consulted sources and explicit unknowns for deterministic policy handling.
Watch API and MCP guide →Compare proposed actions against delegated mandates, trace changed claims across linked reports, record MFA-authenticated sign-off, and export signed Agent Evidence Receipts.
Intelligence API guide →Your application orchestrates execution. Watch and Intelligence Workspace use separate scoped credentials so least-privilege boundaries hold end to end.
Evaluate public addresses and chains before settlement. Handle sanctions matches, partial coverage and source outages explicitly in code.
Surface mint, pause and approval signals where supported. Keep signing keys and final execution strictly inside your own infrastructure.
Supplier claims change after initial approval. Flag every linked decision automatically before an agent renews spend.
Agent Action Review and Ed25519-signed Agent Evidence Receipts are live for early-access teams across 14 network configurations, with independent verification via /.well-known/8thguard-receipt-keys.json. Payment Preflight and RWA Preflight extend this architecture into paid APIs and tokenised real-world assets.
Your agent → Watch
Your agent submits a wallet, transaction or contract and its blockchain. No wallet connection is needed.
Inside Watch
Watch consults the available sources and applies versioned rules. Missing evidence stays visible.
Watch → Your agent
Read the indicators alongside coverage and uncertainty. An empty indicator list does not establish safety.
Your application
Your system can continue, pause or ask a person according to its own policy. Watch provides evidence; it does not execute the transaction.
Request public-chain evidence through an authenticated API. Contact the team for an issued key.
Read the HTTP API guideConnect through the Python source adapter with stdio or Streamable HTTP. Requires an authorised source checkout and engine access.
Read the MCP tool guideThe Watch response lists consulted sources, failures and rule-linked signals. An empty indicator list does not establish safety.
Keep the issued ID and rules version beside your application’s decision. The example below is synthetic and verifies only as a sample.
Report text must not override your agent’s tools or authorisation. Your system handles execution and any escalation to a person.
{
"contract": "1.0",
"id": "8G-S-2026-00005-G",
"sample": true,
"subject": {
"kind": "transaction",
"chain": "evm:1",
"value": "0x3333333333333333333333333333333333333333333333333333333333333333"
},
"tier": "quick",
"issued_at": "2026-10-21T09:00:00Z",
"rules_version": "2026.10.1",
"coverage": [
{
"source": "fixture",
"status": "available"
}
],
"observations": [
{
"rule": "R025",
"text": "Synthetic test evidence only; no live allegation.",
"source": "fixture"
}
],
"indicators": [
{
"code": "unlimited_approval",
"severity": "high",
"rule": "R025"
}
],
"unknowns": [
"Sample record. No live intelligence or safety conclusion."
],
"confidence": "partial",
"next_step": "Run a real check with current source evidence.",
"verify_url": "https://8thguard.com/verify/8G-S-2026-00005-G"
}