Source Trace

AI coding trace & ROI

Running Copilot, Cursor, and Claude? See how much code is AI-generated, how many engineers use agents, and which models are worth the spend.

No workflow changes. Onboard your team in minutes with our VS Code extension.

Claude
Codex Codex
VS Code
Kilo Code
OpenCode OpenCode

Teams: $5/engineer/month or $50/year. Personal: free.

No demos. No sales calls. No credit card to start.

AI coding adoption trend by team

Know how your teams are using AI

One place for engineering leaders to see how teams, agents, and tools work together. Track adoption by team and repository. Measure code rework rate.

Use scorecards to accelerate adoption and improve how engineering runs. Retain the best AI talent and create a culture of experimentation and accountability.

AI code survived and discarded before commit by agent

Compare ROI across agents and models

Most teams run several coding agents and models with no systematic way to compare them. Compare code written, code kept, and rework across languages, teams, and repositories.

Know where each tool works best and where it creates rework. Use data to get buy-in from teams and leadership. Go into renewals with evidence from your own codebase.

AI Git Blame

See which code was generated, even when several models touched one commit. Track agent and model versions, months later.

Powered by Git. Your code stays on your machines. Attribution syncs to your team git, in open format.

Commit Diff View Commit Diff View

Zero config, works with most AI tools

Source Trace VS Code extension tracks changes to your source code - whether they come from you, other extensions, or terminal agents like Claude. Each line is tagged locally on your machine to power AI Blame.

Major AI coding agents and models are supported. More coming soon.

Claude
Claude
Codex Codex
Codex
VS Code
VS Code
Kilo Code
Kilo Code
OpenCode OpenCode
OpenCode

Why Source Trace?

Team Dashboard

Company-wide AI metrics in a shared dashboard. See where agents are used, how much code survives, and which models are worth the budget.

Read the docs →

ROI per model

Which subscriptions are worth renewing? Survival rate by model and language answers it on your codebase, not someone's benchmark.

AI Git Blame

Track who wrote each line - AI or your teammate? Which version of each model was used, even months later.

Privacy First

Attribution happens on each developer's machine. Your code is never uploaded. Only counts, file paths, and hashed identifiers leave the machine.

See what your AI coding tools produce.

Get VS Code Extension