Inspiration

Most developers are using Ai to generate code by a few of them are Instrumenting the Codebase and this takes a While to learn the tools like Jaeger,Docker,Grafana, Prometheus and other Observability tool.

What it does

TraceBeacon is a MultiAgent that is Able to Detect the Framework,Analyze the Codebase for Routes and Apis and Correctly setup openTelemetry for the Users Codebase and Read the telemetry if the user doesn't understand Breakage in their system.

How we built it

I built TraceBeacon with Agent handoff in Order to Keep the Agents Specialized without blotting its context window , so we have different agents to go through the different steps of setting up observability.

Challenges we ran into

Agents downloading wrong/outdated packages and documentation, Agents Not Validating their Code via the setup if there Exists some Code and thus had to setup Multiple Guardrails.

Accomplishments that we're proud of

TraceBeacon is an Ai Agent that has been Able to Detect,Analyze,Implement and Validate throughout the whole setup and Pipeline to Make sure Users Dont have to spend hours Reading and Debuging Observability. TraceBeacon Produces a Well Documented Observability.md that Teaches the users how to Read Observability data Reducing time to Knowledge for the developer on how to use it.

What we learned

Agents Need Guadrails and well Specified Prompt to go through the steps to Make sure the would Work with any different Model Providers.

What's next for TraceBeacon

Improvement on how it implements Observabilty and Integration with the upcoming observability Feature in Gitlab.

Built With

Share this project:

Updates