Skip to content
Devtools

Log Anomaly Explainer

Finds the unusual lines in huge logs and explains them.

Intermediate24h1–3 people15/20 fit
TL;DR

Build Log Anomaly Explainer for backend and on-call engineers in ~24h with 1–3 people. Stack: Python + LLM + React.

Why it wins: Incident-time value with a clear highlight.

Build flow

01 · Problem
During incidents, engineers scroll thousands of log lines looking for the cause.
02 · Solution
Cluster log templates, flag new or spiking patterns, and explain the likely cause.
03 · Build
  1. 1Parse log lines
  2. 2Template mining
  3. 3Baseline vs spike
  4. 4Explanation UI
04 · Demo
Upload a log file → one spiking error highlighted with a plain explanation.

MVP checklist

  • Log upload
  • Template clustering
  • Spike detection
  • Plain explanations

Tech stack

TechRole
PythonBackend / scripting
LLMLanguage model
ReactFrontend UI

Related ideas

More Devtools

  • Devtools

    Incident Postmortem Drafter

    Draft a blameless postmortem from the incident channel and alerts

    Intermediate14h1–2
    Slack APILLMNode.js
    17/20

  • Devtools

    Trace Waterfall Explainer

    Explain why a request was slow from its distributed trace

    Advanced22h1–2
    OpenTelemetryNode.jsLLM
    16/20

  • Devtools

    Flaky Test Detector

    Identifies tests that randomly fail and why

    Intermediate24h1–3
    PythonJUnit XMLReact
    15/20