See how an AI-driven security assessment unfolds, from the first prompt to the evidence.
TraceForge turns recorded model exchanges and tool activity into an interactive, evidence-linked timeline. Follow agents working in parallel, review the conversation, and inspect the inputs and outputs behind key milestones.
The current preview brings together a storyline, chat and terminal views, smart playback pacing, and an evidence inspector. Playback reviews recorded activity; it does not execute the recorded tools.
- Artex is the first integration, with recording support also available for Codex in the development preview.
- Planned: humanwill-jev-policies integration to bring policy decisions and the AI activity behind them into the same review experience. That project's supported integrations include the LiteLLM and Agentgateway AI gateways, plus supported GitHub Copilot hooks. Coverage depends on the connector; the TraceForge adapter is planned.
- Planned: reviewed Python-based verification procedures to help compare behavior before and after a fix.
The Artex storyline, conversation and milestone cards:
A closer look at model exchanges, tool activity and recorded outcomes:
TraceForge is under active development. This repository contains the coming-soon preview; implementation and installation instructions are not yet published.