Phase 1 · x402 · async orchestration

Malware analysis
native to AI agents

An agent submits a file, pays per analysis via x402 / HTTP 402 and gets a structured verdict — no human in the loop. The “VirusTotal” for agents, powered by Assemblyline 4.

▶ View the API Contact us
agent@swarm — analysis
# The agent submits a sample (STATIC tier = synchronous)
$ curl -F "file=@sample.bin" -F "tier=STATIC" \
    https://agentsec.ca/v1/analyses

{ "verdict": "malicious", "score": 920,
  "sha256": "a1b2…f9", "report": "report.v1" }
// The platform

Built for machines, not humans

Programmatic payment, structured output, and a ports & adapters architecture that scales without rewrites.

🤖

Agent-native

MCP + REST delivery. The agent negotiates payment and retrieves a machine-readable verdict, end to end.

💸

x402 payments

Compliant HTTP 402 challenge, USDC settlement via facilitator, replay protection, sanctions screening, and a credit if analysis fails.

🧬

Structured verdict

Versioned report.v1 + STIX contract. Content-addressed by sha256: dedup, idempotency, cost cache.

📊

Observable

2-tier orchestration, Prometheus metrics and a real-time Grafana cockpit dashboard.

// How it works

Three calls, zero friction

From file to verdict, the protocol is designed to be driven by an autonomous agent.

01

Submit

The agent POSTs the file to /v1/analyses. Response 402 with the payment terms.

02

Pay

The agent signs a USDC authorization and replays with the X-PAYMENT header. Settled via facilitator.

03

Retrieve

STATIC → instant verdict. DEEP → 202 + polling/webhook once detonation completes.

// API

Submit a file in one request

Simple REST, idempotent by hash. The cached verdict is the cheapest path.

~/agentsec
# STATIC tier — synchronous, instant report
$ curl -F "file=@sample.bin" -F "tier=STATIC" \
    https://agentsec.ca/v1/analyses

# DEEP tier — asynchronous, detonation then polling
$ curl -F "file=@sample.bin" -F "tier=DEEP" \
    https://agentsec.ca/v1/analyses

# Cached verdict — the cheapest path
$ curl https://agentsec.ca/v1/files/<sha256>/verdict