Retrace

Record, replay, fork & share AI agent executions. See every LLM call, tool call, and error your agent makes, then debug in seconds. Free for 1,000 traces/mo.

Free / paid plans
Visit website ↗
About the product

What does Retrace do?

Retrace records every model and tool call an AI agent makes. When a run fails, you can replay the exact execution, fork from the failing step, change inputs or prompts, and verify the fix before deploying. It also provides runtime guardrails, circuit breakers, evaluation gates for CI/CD, and multi-agent tracing. SDKs for Python and TypeScript.

Who it’s for

Who should use Retrace?

Developers and teams building or operating AI agents are the clearest fit based on how the product is positioned.

Quick fit check

Best fit

Pricing: Free / paid plans

Platforms: Web

Cost

How much does Retrace cost?

Free / paid plans Check the provider’s own page for the latest pricing, packages and terms.

Directory note

Details, pricing and availability can change. We use provider information, verified owner updates and editorial review where available. Check the provider’s own page before making a purchase, booking or important decision.

Profile last updated 10 Nov 2026.
Community ratings

Ratings & reviews

Rate this listing

Sign in to rate this listing and share first-hand feedback.

Sign in or join
No written reviews yet

Be the first to share a useful first-hand impression.

Discussion

Questions and experiences

Used Retrace? Share something that would help another person decide whether it is right for them.

Growth help from the team behind The Tool Directory

Built the product. Need help getting people to find, choose or keep using it?

The Tofu Group helps founders diagnose where growth is actually getting stuck — search and AI visibility, positioning, conversion, content, acquisition or retention — before adding more marketing for the sake of it.