Enterprise agent infrastructure

Agent Payment Ledger

A payment-aware ledger for autonomous agents. Inspect tool calls, approval gates, retries, errors, x402 payments, and final outcomes in one audit trail.

3seeded vendor-risk runs
6agent event types
1Splunk-ready export path
Agent Payment Ledger dashboard showing agent run timelines and risk flags

What judges can inspect

The demo is not another chat interface. It is operational evidence for teams asking what an autonomous workflow touched, spent, retried, blocked, and finished.

success

Vendor risk review

Records search, human approval, x402 enrichment payment, and final cleared outcome.

retry

Supplier onboarding

Flaky tool call, retry evidence, recovered workflow, and spend attribution by agent.

blocked

High-risk payment

Approval request, rejection, blocked paid action, final error state, and exportable JSON.

Small SDK surface

Agent Payment Ledger works around existing agent runtimes. Wrap the workflow, emit events, and keep the agent stack you already use.

const run_id = await startRun({
  agent,
  workflow
});

await trackToolCall({
  run_id,
  step_name,
  tool_name,
  fn
});

await trackApproval({
  run_id,
  approval_status
});

await trackPayment({
  run_id,
  endpoint,
  cost,
  currency
});

await finishRun({
  run_id,
  status,
  summary
});
Designed for NandaHack-style infrastructure evaluation and Splunk Agentic Ops telemetry. The Splunk HEC adapter in adapters/splunk-hec exports run events as agentops:run_event.

Submission assets

Everything needed to judge, run, inspect, and adapt the project is public.