← Back to Discover

Source Trace

VS Code extension and team dashboard that tag which lines of code came from AI agents and models, and compare code written, kept and reworked across tools, teams and repositories.

Observability & evals
Visit website ↗

What it does

Source Trace tracks changes to your source code, whether they come from you, other extensions or terminal agents, and tags each line locally on the developer's machine to power what it calls AI Blame. It records which agent and model version produced code, including when several models touched one commit, and keeps that information available months later. A team dashboard aggregates adoption by team and repository, rework rate and how much AI-written code survives, so engineering leaders can compare models and subscriptions using their own codebase. The page names Claude, Codex, VS Code, Kilo Code and OpenCode among supported tools and says more are coming. The site states that attribution happens on each developer's machine and that code is not uploaded. Only counts, file paths and hashed identifiers leave the machine, and attribution syncs to the team's git in an open format.

Capabilities

  • Line-level attribution of AI-written code, described as AI git blame
  • Tracking of agent and model versions over time
  • Team dashboard with adoption by team and repository
  • Survival and rework rates by model and language
  • Local tagging with no code upload
  • Zero-config VS Code extension

When to use it

  • Deciding which AI coding subscriptions are worth renewing
  • Measuring AI adoption across engineering teams
  • Seeing which parts of a codebase were AI-generated