helix
Analyze a trace

AI AGENT COST & RISK SCANNER

See where your agents waste money.

Helix scans agent traces for duplicate model calls, bloated prompts, tool loops, failed steps, and exposed credentials. Then it shows what to fix.

Analyze a trace
helix.trevyn.dev/r/demo-trace

Demo trace · cost and risk report

Wasteful support agent

OpenAI Agents
Well Done

“Your agent paid twice for the same thought.”

1 critical · 2 warning · 0 notice

  1. SEV 2cost
    duplicate llm call
    $2.17 waste

Example report · demo trace

What costs money. What can break.

Deterministic cost, reliability, and security checks for completed agent traces.

  • Duplicate LLM calls and repeated 2,000+ token prompt prefixes
  • Single calls with more than 20k input tokens
  • The same tool called 4+ times with identical arguments
  • Failed tools with no later retry
  • Steps that run longer than 15 seconds
  • Traces that end in an unhandled error
  • API keys, bearer tokens, and private keys in trace data
  • Emails, phone numbers, and plain-HTTP tool URLs