DCL Evaluator

Endpoints (19 live)

• POST /evaluate/fast — Low-latency pre-action screening for routine agent responses and tool-call decisions. (0.01 USDC on Base)
• POST /evaluate/strict — Higher-confidence evaluation for high-impact, sensitive, or difficult-to-reverse agent actions. (0.05 USDC on Base)
• POST /evaluate/secrets — Scan text for exposed credentials, API keys, private keys, database URLs, and webhook secrets. COMMIT when nothing matches. NO_COMMIT on any finding. POST JSON {"response","agent_id"}. (0.02 USDC on Base)
• GET /audit/:tx_hash — Read one audit-chain record by tx_hash. GET path parameter, no JSON body. Returns verdict, confidence, agent_id, and reason. Does not return raw content. (0.1 USDC on Base)
• POST /evaluate/batch — Evaluate a list of texts in one paid call. Each item may name a built-in policy. The call is rejected when the list is longer than max_items (default 20). POST JSON {"items","agent_id","max_items"}. (0.1 USDC on Base)
• POST /evaluate/pii — Scan text for personal data: email, phone, national id, bank card, IBAN, crypto address, IP address, passport. COMMIT when nothing matches. NO_COMMIT on any finding. POST JSON {"response","agent_id"}. (0.02 USDC on Base)
• POST /evaluate/jailbreak — Detect jailbreaks, prompt injection, and instruction conflicts before the agent follows them. (0.02 USDC on Base)
• POST /evaluate/quality — Check agent outputs for quality issues, inconsistency, and behavioral drift. (0.03 USDC on Base)
• POST /evaluate/safety — DCL Trust Oracle. Call only when the text must include the substring AI and must not include "I cannot be held responsible" or "no guarantees". POST JSON {"response","agent_id"}. COMMIT means those checks passed and the action may proceed. NO_COMMIT means do not execute the checked action. Not a tool-call guardrail, prompt-injection detector, or jailbreak screen; use /evaluate/fast for the phrase check. (0.01 USDC on Base)
• GET /evaluate/secrets — Scan text for exposed credentials, API keys, private keys, database URLs, and webhook secrets. COMMIT when nothing matches. NO_COMMIT on any finding. POST JSON {"response","agent_id"}. (0.02 USDC on Base)
• GET /audit/1 — Read one audit-chain record by tx_hash. GET path parameter, no JSON body. Returns verdict, confidence, agent_id, and reason. Does not return raw content. (0.1 USDC on Base)
• GET /audit/1/deep — Read one audit-chain record by tx_hash, plus drift context and tamper location. GET path parameter, no JSON body. (0.5 USDC on Base)
• GET /evaluate/batch — Evaluate a list of texts in one paid call. Each item may name a built-in policy. The call is rejected when the list is longer than max_items (default 20). POST JSON {"items","agent_id","max_items"}. (0.1 USDC on Base)
• GET /evaluate/pii — Scan text for personal data: email, phone, national id, bank card, IBAN, crypto address, IP address, passport. COMMIT when nothing matches. NO_COMMIT on any finding. POST JSON {"response","agent_id"}. (0.02 USDC on Base)
• GET /evaluate/safety — Baseline safety screening for harmful, unsafe, or policy-violating agent outputs. (0.01 USDC on Base)
• GET /evaluate/jailbreak — Detect jailbreaks, prompt injection, and instruction conflicts before the agent follows them. (0.02 USDC on Base)
• GET /evaluate/fast — Low-latency pre-action screening for routine agent responses and tool-call decisions. (0.01 USDC on Base)
• GET /evaluate/strict — Higher-confidence evaluation for high-impact, sensitive, or difficult-to-reverse agent actions. (0.05 USDC on Base)
• GET /evaluate/quality — Check agent outputs for quality issues, inconsistency, and behavioral drift. (0.03 USDC on Base)