FailTrace

FailTrace is a webhook reliability and replay-decision dashboard built for operators who need more than raw delivery logs.

Instead of only showing failed webhook attempts, FailTrace classifies delivery state, identifies likely failure reasons, estimates replay confidence, scores endpoint health, and recommends the next operational action.

Live Demo

• Frontend: https://fail-trace.vercel.app
• Backend API: https://failtrace.onrender.com
• API docs: https://failtrace.onrender.com/docs

What It Does

• Reliability dashboard for webhook delivery health
• Event analyzer for replay-safety decisions
• Endpoint monitor with health trends and event history
• Failure classification for signatures, timeouts, rate limits, deleted endpoints, malformed responses, and duplicates
• Replay confidence scoring with recommended actions
• LangGraph workflow for the replay-decision pipeline
• ML-assisted single-event analysis with deterministic guardrails

Product Flow

Dashboard

The main dashboard gives a fast operational view of webhook reliability:

• total observed events
• failed deliveries
• average endpoint health
• unsafe replay count
• recent event table
• retry distribution
• endpoint health chart

Event Analyzer

The analyzer accepts an event ID and returns:

• delivery state
• failure reason
• safe-to-replay verdict
• recommended action
• endpoint health score
• replay confidence score
• plain-language explanation

Endpoint Monitor

The endpoint view helps identify whether failures are isolated to one event or part of a broader endpoint issue:

• endpoint health score
• total events
• failure rate
• last seen timestamp
• score trend
• event history drawer

Architecture

Next.js frontend
        |
        | HTTP API
        v
FastAPI backend
        |
        | LangGraph decision workflow
        v
SQLite + generated webhook dataset
        |
        | optional single-event inference
        v
ML replay intelligence layer

Tech Stack

• Frontend: Next.js, React, TypeScript, Tailwind CSS, Recharts, Axios
• Backend: Python, FastAPI, Pandas, SQLite, Pydantic, LangGraph
• ML: scikit-learn, NumPy, joblib
• Deployment: Vercel frontend, Render backend

Backend API

Method	Route	Purpose
GET	/health	API health check
POST	/seed	Generate and load webhook sample data
GET	/events	List dashboard-ready event summaries
GET	/events/{event_id}	Fetch a single event summary
GET	/endpoints	List endpoint health monitors
POST	/analyze	Analyze an event or webhook attempt
GET	/attempts	List raw delivery attempts
GET	/decisions	List analyzed decision rows
GET	/analytics/summary	Aggregate reliability metrics

Deployment

Frontend

The frontend is deployed on Vercel from the frontend directory.

Required environment variable:

NEXT_PUBLIC_API_BASE_URL=https://failtrace.onrender.com

Backend

The backend is deployed on Render.

Build command:

pip install -r requirements.txt

Start command:

uvicorn api.main:app --host 0.0.0.0 --port $PORT

Recommended environment variable:

PYTHON_VERSION=3.12.8

Run Locally

Backend

python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
python scripts/generate_dataset.py
uvicorn api.main:app --reload

Backend URLs:

http://127.0.0.1:8000/health
http://127.0.0.1:8000/docs

Frontend

cd frontend
npm install
npm run dev

Frontend URL:

http://localhost:3000

For a local backend, use:

NEXT_PUBLIC_API_BASE_URL=http://localhost:8000

Demo Script

1. Open the dashboard and show reliability metrics, recent events, retry distribution, and endpoint health.
2. Open the analyzer and submit an event ID such as evt_pay_91k2.
3. Explain the safe-to-replay verdict, failure reason, confidence score, and recommended action.
4. Open endpoints and inspect an endpoint history drawer.
5. Close by explaining that FailTrace turns webhook delivery logs into replay decisions.

Status

FailTrace is a hackathon-ready full-stack demo with a deployed frontend, deployed FastAPI backend, live API contract, and ML-assisted replay intelligence path.