Hugging Face's logo

Spaces:
Tuathe
/
watchtowerai-log-analyzer
Sleeping

App Files Community
watchtowerai-log-analyzer / README.md
Tuathe's picture
Tuathe
Update README.md
17a77b6 verified
preview code
|
raw
history blame contribute delete
1.62 kB

A newer version of the Streamlit SDK is available: 1.49.1

Upgrade
metadata
title: WatchTowerAI - Log Analyzer
emoji: 🛰️
colorFrom: purple
colorTo: pink
sdk: streamlit
sdk_version: 1.32.0
app_file: streamlit_app.py
pinned: false

WatchTowerAI – Log Classification & Automated Runbook Generator

A production-grade AI tool to analyze logs in real-time, classify them using a fine-tuned Flan-T5 model, generate markdown runbooks, and automatically alert your team via Slack — all deployed on Hugging Face Spaces.

Features

  • LLM-powered log classification (INFO, WARNING, ERROR, CRITICAL, SECURITY)
  • Runbook generation using Flan-T5
  • Upload runbooks to AWS S3
  • Send real-time Slack alerts
  • Hybrid fallback using rule-based logic if model is uncertain
  • Streamlit Web UI + .log file upload support

Tech Stack

  • LLM: Fine-tuned Flan-T5
  • Frontend: Streamlit
  • Cloud: AWS S3
  • Alerts: Slack Webhook
  • Deployment: Hugging Face Spaces

Screenshot

screenshot

How to Use

  1. Paste or upload a .log file
  2. Click “Classify + Generate Runbook”
  3. Download the markdown or view it in S3
  4. Slack alerts will be sent automatically

Directory

  • streamlit_app.py – UI + logic
  • upload_to_s3.py – S3 upload function
  • notify_slack.py – Slack integration
  • runbooks/ – Markdown output folder
  • codementor-flan/ – Your fine-tuned model folder

AI Model

Fine-tuned on domain-specific QA + system logs using Flan-T5 for robust multi-class log categorization and response generation.

Made by Chetan