Every step an agent takes, in order
A trace in Definable is the record of one agent run: the model the router picked and why, then every model call and tool call in order, with inputs, outputs, tokens, cost and timing. Every chat and scheduled run keeps its trace, so any answer can be explained after the fact.
How traces work
Every run keeps its trace: the router's pick of model, each model call and each tool call, in order and with their timings, so any answer can be explained after the fact.
At a glance
- Tree and waterfall
- Formatted or JSON
- Tokens and cost per call
- Runs, sessions and errors
- 01
Every run, newest first
Chats and scheduled tasks land in one list with their status, duration and cost. Filter by source or status, or search by name or ID.
- 02
Each step, as a tree or a waterfall
The router's pick, every model call in blue and every tool call, drawn against time so the slow step stands out.
- 03
Open any step
See what went in and what came out, formatted or as JSON, with its tokens, time to first token and cost.
Traces FAQ
What does a trace show?
Every model call and tool call in order, with inputs, outputs, tokens, cost and timing, which model the router picked and why, and the ratings people gave the answer.
How do fast and slow models work?
Give an agent a fast model beside its main one. For each new message a small router decides which should answer, and either model can hand the work to the other mid-run. The router is free, and its choice shows in the trace.
Read how it worksAre my provider keys safe?
Keys are encrypted with AES-256-GCM and never appear in logs or traces.
How do I know an agent is ready?
Write eval cases with checks, including an LLM judge, and compare every change against a baseline. Evals run on the draft before you publish, so you see anything that got worse first.