AI AGENT SKILLS // AUTONOMOUS DUE DILIGENCE
Callable intelligence skills for autonomous AI workflows.
Equip your AI agents with 9 discrete compliance tools. Query registries, screen sanctions, trace crypto counterparty risk, and evaluate fraud probability with zero human middleware.
The Autonomous Intelligence Loop
Autonomous AI agents fail when APIs return unvalidated HTML or generic error strings. Diliguard provides fully typed JSON schemas that agents can discover, execute, and validate deterministically.
1. Tool Discovery
Your agent reads tool definitions via Model Context Protocol or OpenAI function calling specifications, discovering input constraints automatically.
2. Execution & Resolution
The agent issues structured parameters. Diliguard queries authoritative government databases, sanctions registries, and blockchain nodes in parallel.
3. Deterministic Decisions
Returns structured risk scores and audit trails, allowing your agent to branch logic into automatic approval, enhanced screening, or human escalation.
Directory of 9 Callable Skills
Each tool is callable directly via HTTP API, Model Context Protocol server, or LangChain and AutoGen tool bindings.
run_corporate_kyc: Corporate KYB & Registry Unwrapping
Executes full corporate entity verification, ultimate beneficial ownership (UBO) unwrapping across 16 primary jurisdictions, LEI validation, and multi-source adverse media queries.
INPUT PARAMETERS
| Parameter | Type | Required | Description |
|---|---|---|---|
company_name | string | Yes | Legal registered entity name |
jurisdiction | string | No | Two-letter ISO country code (e.g. GB, US, DE, SG) |
registration_number | string | No | Official government registry company number |
SAMPLE STRUCTURED RESPONSE
{
"status": "active",
"company_number": "12345678",
"incorporation_date": "2018-04-12",
"registered_address": "100 Bishopsgate, London EC2N 4AG, UK",
"ubo_structure": [
{"name": "Arthur Pendelton", "ownership_pct": 65.0, "sanctions_match": false}
],
"risk_score": 12,
"risk_tier": "LOW"
}run_person_trace: Executive Person Trace & OSINT
Conducts deep background screening on individual officers, directors, and beneficial owners across global PEP databases, Interpol red notices, court judgments, and digital footprint telemetry.
INPUT PARAMETERS
| Parameter | Type | Required | Description |
|---|---|---|---|
full_name | string | Yes | Legal first and last name of the subject |
dob | string | No | Date of birth in YYYY-MM-DD format |
country | string | No | Country of residence or primary nationality |
email | string | No | Corporate or personal email address for OSINT correlation |
SAMPLE STRUCTURED RESPONSE
{
"full_name": "Marcus Vance",
"pep_status": false,
"sanctions_hits": 0,
"adverse_media_count": 0,
"directorships_count": 3,
"associated_entities": ["Apex Logistics Ltd", "Vance Capital Partners"],
"risk_score": 8,
"risk_tier": "LOW"
}run_crypto_investigation: Crypto Digital Asset Investigation
Traces blockchain transaction graphs across Bitcoin, Ethereum, Solana, and EVM chains. Flags interactions with sanctioned wallets, mixers, illicit bridges, and high-risk counterparties.
INPUT PARAMETERS
| Parameter | Type | Required | Description |
|---|---|---|---|
wallet_address | string | Yes | Public blockchain address or transaction hash |
blockchain | string | No | Network identifier (e.g. eth, btc, sol, arbitrum, bsc) |
SAMPLE STRUCTURED RESPONSE
{
"address": "0x71C...3a9",
"blockchain": "ethereum",
"total_received_usd": 482000.00,
"mixer_exposure_pct": 0.0,
"sanctioned_counterparty_hits": 0,
"direct_cluster_risk": "CLEAN",
"composite_risk_score": 15
}run_vendor_risk: Third-Party Vendor Risk Auditing
Automates third-party supplier diligence, validating business operational status, financial solvency signals, pending litigation records, and regulatory enforcement actions.
INPUT PARAMETERS
| Parameter | Type | Required | Description |
|---|---|---|---|
vendor_name | string | Yes | Legal name of the vendor or supplier |
domain | string | No | Corporate website URL or root domain |
tax_id | string | No | VAT, EIN, or national tax identifier |
SAMPLE STRUCTURED RESPONSE
{
"vendor_name": "CloudScale Systems Inc",
"operational_status": "active",
"credit_rating_proxy": "AAA",
"litigation_records": 0,
"regulatory_actions": 0,
"vendor_risk_tier": "TIER_1_APPROVED"
}run_ultimate_fraud: Composite Fraud Prevention Analysis
Runs comprehensive multi-vector synthetic identity detection, combining corporate registry telemetry, domain age, device fingerprinting, and behavioral risk flags.
INPUT PARAMETERS
| Parameter | Type | Required | Description |
|---|---|---|---|
entity_name | string | No | Business name under evaluation |
contact_email | string | No | Primary applicant contact email |
ip_address | string | No | IP address of the session |
device_fingerprint | string | No | Client hardware fingerprint hash |
SAMPLE STRUCTURED RESPONSE
{
"fraud_score": 4,
"synthetic_identity_probability": 0.02,
"disposable_domain": false,
"ip_risk_level": "LOW",
"decision": "APPROVE",
"confidence_score": 0.98
}run_aml_pep_screening: AML & Institutional PEP Screening
Screens entities and individuals against 140+ real-time global sanctions lists including OFAC, EU Consolidated, UK HMT, UN Security Council, and FATF high-risk registries.
INPUT PARAMETERS
| Parameter | Type | Required | Description |
|---|---|---|---|
query | string | Yes | Individual or legal entity name to screen |
fuzzy_threshold | number | No | Fuzzy matching threshold between 0.70 and 1.00 (default: 0.85) |
SAMPLE STRUCTURED RESPONSE
{
"query": "Novikov Holding",
"matches_found": 0,
"databases_checked": 142,
"sanctions_status": "CLEAR",
"pep_status": "CLEAR",
"screening_timestamp": "2026-10-05T20:15:00Z"
}run_registry_check: Direct Government Registry Check
Queries national statutory corporate registries directly to retrieve raw official registration certificates, officer rosters, share capital structures, and annual return filings.
INPUT PARAMETERS
| Parameter | Type | Required | Description |
|---|---|---|---|
company_number | string | Yes | Official registry filing number |
country_code | string | Yes | Two-letter ISO country code (e.g. GB, DE, FR, NL, US) |
SAMPLE STRUCTURED RESPONSE
{
"registry": "Companies House UK",
"company_number": "08912345",
"status": "Active",
"type": "Private Limited Company",
"next_accounts_due": "2027-01-31",
"active_directors": 2
}run_ip_intelligence: IP & Network Infrastructure Telemetry
Analyzes connection IP addresses for VPN usage, commercial proxies, Tor exit nodes, hosting datacenter allocation, and residential proxy networks.
INPUT PARAMETERS
| Parameter | Type | Required | Description |
|---|---|---|---|
ip_address | string | Yes | IPv4 or IPv6 address to inspect |
SAMPLE STRUCTURED RESPONSE
{
"ip": "194.26.29.112",
"is_proxy": false,
"is_vpn": false,
"is_tor": false,
"is_datacenter": false,
"asn": "AS2856",
"isp": "British Telecommunications",
"country": "GB"
}run_transaction_scoring: Transaction Scoring & Velocity Risk
Calculates real-time fraud probability scores and velocity spikes on monetary transactions, flagging structuring patterns and suspicious counterparty flows.
INPUT PARAMETERS
| Parameter | Type | Required | Description |
|---|---|---|---|
amount | number | Yes | Transaction amount in specified currency |
currency | string | Yes | Three-letter ISO currency code (e.g. USD, EUR, GBP) |
sender_id | string | Yes | Unique identifier of the sending party |
recipient_id | string | Yes | Unique identifier of the receiving party |
SAMPLE STRUCTURED RESPONSE
{
"transaction_id": "tx_9941a8b",
"risk_score": 18,
"recommendation": "ALLOW",
"velocity_24h_count": 3,
"unusual_amount_flag": false,
"structuring_probability": 0.01
}Multi-Agent Orchestration Patterns
Coordinate multiple specialized agents to solve end-to-end compliance workflows autonomously.
Pattern 1: Autonomous Merchant Onboarding
Workflow: Agent runs run_corporate_kyc on the registered company, extracts all named directors, executes run_person_trace on each individual, and scores the opening transaction via run_transaction_scoring.
Pattern 2: Supply Chain Vendor Screening
Workflow: Procurement agent triggers run_vendor_risk for corporate solvency, verifies technical infrastructure with run_ip_intelligence, and validates ultimate beneficial owners against run_aml_pep_screening.
Security & Zero Data Retention for Agent Loops
No Model Training
Customer payload queries, target identifiers, and verification artifacts are never used to train machine learning models.
Client Environment Secrets
API keys stay inside your local agent runtime environment variables. MCP servers communicate over authenticated stdio or TLS pipes.
Equip your AI agents with Diliguard skills today.
Get your API credentials immediately or explore the complete developer integration guides.