Hacker

Dhanraj Avhad

@dhanraj30

Joined Oct 2026Fullstack software developer at Z50 evolutionIndia

Projects

  • Top 50

    TraceLens: the incident agent that rules out the wrong suspects

    Graph HacksOct 2026

    TraceLens is an AI incident investigator built on FalkorDB. When production breaks, it walks a graph of your system upstream from the failing service, treats every recent deployment as a suspect, and tries to disprove each one with five graph checks. It reports the best-supported root cause with the exact path and records that prove it, the suspects it ruled out and why, a matching past incident, and a runbook-based fix that a human approves. Everything it learns is written back into the graph, so the next investigation is faster. Outside AI agents can use the same tools over MCP; a LangGraph agent picks its own tools, and the server re-verifies its conclusions before saving them. Finding the root cause of a production incident means connecting many records: what depends on what, what changed and when, where errors started first, and whether this happened before. Engineers stitch this together by hand under pressure and often blame the most recent deploy to the broken service, even when the real cause is a change two services away. AI chatbots don't fix this: they retrieve text and guess at the connections, and they can't show how they reached an answer. TraceLens is for on-call engineers, SREs and platform teams who need a fast answer they can verify. Every conclusion comes with its graph path and evidence, wrong suspects are shown as ruled out with proof, and when the evidence is missing it says "inconclusive" instead of inventing a cause.