Show HN: We made a VS Code extension to recreate a debugger experience from logs

[arthurgousset]

A month ago [1], we made an MCP server so Cursor can debug Node.js on its own. We emailed every person that starred our repository [2] and learnt that frontend devs really want to give Cursor access to browser logs, and that backend devs (our intended audience) do not use debuggers nearly as much as we thought.

We interviewed friends across startups and discovered that they use logs to debug, because they can’t run services locally on their machine. The services (1) require too much disk, RAM, or CPUs to run locally, (2) have too many service dependencies (think microservices), or (3) are a faff to instantiate locally with a debugger. Instead, our friends instrument their services, deploy them to staging environments via Kubernetes, and then query the logs via data stores (think Grafana, Axiom.co, Google Cloud Logging, etc) or directly (think Kubernetes logs).

We thought: "What if we could recreate a debugger-like experience from logs?". That would save them from browsing logs and trying to make sense of them outside the context of the code base.

We looked into it and made a VS code extension that lets you (1) import logs, (2) go to the line of code associated with a log, and navigate up/down the probable call stack associated with a log.

It's a prototype, but if you're interested in trying it out, we'd love some feedback!

GitHub: github.com/hyperdrive-eng/traceback

---

References:

[1]: https://news.ycombinator.com/item?id=43446659

[2]: 140 Github stars, 69 emails sent (the rest were bots), 19 responses received (= 28% response conversion), 4 meetings held (= 21% meeting conversion).