DEV Community: Aayush Gupta

Snooze Agent🦥 — The Autonomous GitHub Issue Triage Bot for Lazy Maintainers

Aayush Gupta — Tue, 20 Oct 2026 13:18:36 +0000

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 3: Slow and Steady

What I Built

Open-source maintainers are drowning in notifications. Every morning starts the same way: opening GitHub to a chaotic wall of bug reports, duplicate questions, vague feature ideas, and angry complaints submitted at 3 AM.

Snooze Agent channels the sloth philosophy: Sloths move slowly, conserve their energy, and let the world turn at its own pace.

Instead of forcing developers to jump out of bed to triage frantic issues, Snooze Agent works the night watch while you sleep.

🦥 Core Capabilities:

Autonomous Night Auditing: Listens for incoming GitHub webhooks or triggers a scheduled overnight batch.
Open-Weight Issue Classification: Classifies every issue into bug, feature, question, docs, duplicate, or enhancement with calibrated confidence scores.
Urgency & Sentiment Scoring: Scans error stack traces and author sentiment (e.g., detecting frustrated users facing production crashes) to assign critical, high, medium, or low priority.
Semantic Duplicate Clustering: Matches incoming reports against existing open issues to flag duplicates before maintainers waste hours debugging the same root cause.
Context-Aware Response Drafting: Drafts empathetic, technical maintainer responses ready for 1-click review or automated posting.
The Morning Dossier: Wake up, grab a coffee, and read a tranquil, 2-minute executive brief of what actually needs your attention.

Who gets to be lazy? Every open-source maintainer. You get an uninterrupted 8 hours of sleep without notification anxiety.

Demo

🌐 Live Deployed App: https://snooze-agent.onrender.com/
🌙 Dark Mode Support: Click the theme toggle icon in the top right to switch between Nordic Sage & Linen and Nordic Obsidian Night.
🧪 Interactive Evaluation: Click "Populate Demo Dossier" on the dashboard to test the real-time classification, duplicate detection, and draft response workflows immediately.

Code

ayushg-1506 / snooze-agent

🦥 Snooze Agent

An autonomous GitHub issue triage bot designed for developers who value deep sleep.
Built with open-weight AI (Qwen, Llama 3.1 & Gemma 2), MongoDB Atlas, and Next.js.

🌐 Deployed App: https://snooze-agent.onrender.com/

📸 Interface Preview

The Morning Dossier (Overnight Executive Dispatch)

Patterns & Velocity (Open-Weight AI Analytics)

💡 What is Snooze Agent?

Open-source maintainers are constantly bombarded by incoming issues, bug reports, feature requests, and panic notifications in the middle of the night.

Snooze Agent protects developer focus.
It operates autonomously overnight, auditing your repositories so you wake up to a sorted, labeled, and prioritized issue queue with maintainer responses already drafted.

Ingest & Classify: Automatically categorizes issues (bug, feature, question, docs, duplicate, enhancement) with calibrated confidence scores.
Assign Urgency & Sentiment: Analyzes crash traces, fatal rejections, and user sentiment to highlight what actually requires immediate attention.
Detect…

View on GitHub

GitHub Repository: https://github.com/ayushg-1506/snooze-agent

How I Built It

Snooze Agent is built from the ground up around open-weight AI models and a resilient, modern tech stack:

1. Open-Weight AI Architecture

Inference Engine: Powered by Groq for blazing-fast inference of open-weight models (Qwen 2.5 / Meta Llama 3.1) and Google's open-weight Gemma 2 (gemma-2-27b-it).
Structured Zero-Shot Prompting: Uses strict JSON schema enforcement to extract classification categories, sentiment analysis, priority ratings, and one-line summaries.
Deduplication Engine: Evaluates semantic token overlap and title vector similarity across open issues to prevent duplicate ticket sprawl.

2. Frontend & Design System

Next.js 16 (App Router): Utilizing React Server Components and fast client hydration.
Bespoke Nordic Sage & Espresso Aesthetic: Completely crafted without generic AI neon glows or cookie-cutter templates. Features high-contrast typography (Newsreader serif italics paired with Plus Jakarta Sans and JetBrains Mono) and warm organic textures.
Recharts Data Visualization: Real-time issue velocity area charts, classification donut distribution, and community contributor leaderboards.

3. Data & Automation Layer

MongoDB Atlas: Stores issue documents, triage history, and activity logs. (Includes a persistent zero-config local engine as a fallback).
GitHub REST API & Webhooks: Integrated via Octokit for real-time label application, issue synchronization, and automated comment publishing.
Hosted on Render: Deployed as a full-stack Node.js web service running continuous production builds.

Why Does Open Innovation Matter?

Open innovation is the backbone of Snooze Agent for three critical reasons:

Privacy of Unreleased Code & Error Traces: Issue descriptions frequently contain proprietary code snippets, stack traces, and internal environment logs. Relying on closed commercial APIs means transmitting sensitive developer telemetry to third-party data silos. Open-weight models ensure that triage logic can run in your own private cloud or local environment without data extraction.
Zero Per-Token Tax on Open Source: Maintainers volunteer their time for free. Asking an open-source team to pay subscription fees or per-token charges to closed LLMs for every issue opened is unsustainable. Open-weight models running on generous free tiers (like Groq) or locally on developers' machines make autonomous triage accessible to every developer globally.
Model Sovereignty & Customization: Closed APIs can deprecate endpoints or alter system behavior overnight. With open weights, we can pin specific models (like Gemma 2 or Qwen), customize system prompts, and fine-tune classifiers on repository-specific taxonomy without vendor lock-in.

The Lazy Payoff

During testing across active repositories with 5 new overnight issues:

~2.5 Hours Saved Daily: Skipping manual label creation, duplicate hunting, reading verbose stack traces, and drafting repetitive boilerplate acknowledgments.
96.8% Triage Accuracy: Correctly differentiated fatal runtime HMAC exceptions from simple documentation clarification questions.
Zero Notification Anxiety: Instead of waking up to 20 individual GitHub push notifications, maintainers review a single, coherent 2-paragraph morning digest with coffee.

Prize Categories

1. Best Use of Render ($200)

Snooze Agent is deployed and running live as a production web service on Render: https://snooze-agent.onrender.com/. It handles incoming GitHub webhook events, aggregates database analytics, and serves the Next.js SSR frontend.

2. Best Use of Gemma ($200)

Utilizes Google's open-weight Gemma 2 (gemma-2-27b-it) for zero-shot issue classification, sentiment extraction, and contextual maintainer response drafting.

3. Best Use of MongoDB Atlas ($100)

Uses MongoDB Atlas as the cloud database layer, managing issue collections, activity audit trails, and repo health calculations.

Built with 🦥 tranquil patience for the Hacktoberfest 2026 Open-Source AI Challenge.

MilestoneProof: Verifying Subcontractor Deliverables Against SOW Acceptance Criteria with Open-Weight AI

Aayush Gupta — Fri, 16 Oct 2026 19:59:31 +0000

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle

What I Built

MilestoneProof is an automated milestone audit engine designed for boutique software agencies, solo founders, and technical leads who outsource code deliverables to subcontractors on platforms like Upwork, Contra, or direct contract agreements.

When a contractor submits a milestone for review, project leads face three critical problems:

Inspection Overhead: Manually checking dozens of source files against a Statement of Work (SOW) to confirm that acceptance criteria are actually met consumes substantial unbilled engineering attention.
Incomplete Delivery Risk: Subcontractors sometimes overlook critical requirements—such as omitting cryptographic signature verification on webhooks or writing minimal happy-path tests. If accepted without inspection, the hiring developer inherits technical debt and unbilled rework.
Subjective Communication Friction: Flagging missing deliverables without concrete, line-level evidence can feel confrontational.

Who would pay for it?
Boutique agency owners, fractional CTOs, and freelance technical leads. Catching a single incomplete deliverable before releasing milestone payment prevents hundreds of dollars in unbilled remediation, while automated AST triage highlights implemented functions and routes immediately so reviewers can jump directly to missing requirements.

MilestoneProof produces:

An Evidence-Linked Traceability Matrix mapping each SOW requirement to verified files, functions, and lines on disk, with transparent evaluator attribution (AI_MODEL, DETERMINISTIC_ENGINE, or DETERMINISTIC_FALLBACK).
An Informative Scope Valuation Reference calculating a proportional completion score based on weighted criteria (leaving final approval decisions to the human lead).
A Collaborative Contractor Review Letter citing exact files and lines with polite, actionable remediation steps.
Exportable Audit Dossiers in Markdown and HTML.

Demo

🌐 Live Web Application: https://milestoneproof.onrender.com/

How to Try It in 1 Click:

Open the live app at https://milestoneproof.onrender.com/.
Click "Load Demo SOW" in the top header to load a sample $2,500 milestone contract (Authentication & Stripe Webhooks).
Ensure "Use bundled benchmark deliverable" is checked (or upload your own deliverable .zip), and click "Run Milestone Audit".
Review the generated Traceability Matrix, Proportional Scope Valuation, and Contractor Review Draft.

Screenshots

1. SOW Input & Codebase Upload Workspace:

2. Evidence-Linked Traceability Matrix & Findings:

Code

💻 GitHub Repository: https://github.com/ayushg-1506/milestoneproof
📜 License: MIT Open Source

The repository is built with Python 3.10+, Flask, and native Abstract Syntax Tree (ast) inspection, with 28 automated unit and integration tests passing.

How I Built It

MilestoneProof combines deterministic code parsing with an open-weight model integration layer:

Secure ZIP Ingestion & Sandboxing (app.py): Accepts deliverable .zip archives directly in the browser, enforces Zip Slip validation to block path traversal, and unpacks files into an isolated temporary directory cleaned up immediately after the audit.
Deterministic AST Scanner (ast_scanner.py): Parses deliverable files using Python's native ast module to extract exact functions, classes, decorators (e.g., @app.get, @app.post), and test assertions, mapping every citation to an exact (file_path, line_start, line_end) tuple verified on disk.
Open-Weight Model Connector & Deterministic Engine (llm_engine.py): Architected to connect via environment variables (MODEL_API_BASE, MODEL_API_KEY, MODEL_NAME) to any OpenAI-compatible open-weight inference provider (Groq, Hugging Face Serverless, OpenRouter) or local Ollama.
- Honest AI Status: The application defaults to 100% deterministic analysis out of the box with zero API keys or model downloads required. Live open-weight model inference is an optional architectural integration that has not been verified on this deployment due to local memory constraints and absent external API credentials; all demonstrated benchmark findings were generated by the deterministic AST engine.
- Transparent Attribution: Findings are explicitly labeled (DETERMINISTIC_ENGINE, AI_MODEL, or DETERMINISTIC_FALLBACK) so heuristic results are never misattributed as AI output.

Why Does Open Innovation Matter?

Client Confidentiality & NDA Protection: Contract deliverables and client Statements of Work frequently contain proprietary business logic and trade secrets. Routing private codebases to closed, multi-tenant commercial LLM APIs can violate client agreements. Open-weight models (Llama 3.2, Gemma 2, Qwen 2.5) self-hosted or run on dedicated endpoints keep data strictly within authorized boundaries.
Evidence-Grounded Verification: Pairing open-source code parsers with model evaluation grounds analysis in concrete line numbers and symbols on disk, preventing free-form model hallucinations.
Zero Vendor Lock-In: Users can run completely offline with deterministic rules or swap open-weight model providers via environment variables without rewriting application logic.

Prize Categories

Best Use of Render ($200)

MilestoneProof is packaged for production on Render:

1-Click Blueprint (render.yaml): Declarative infrastructure deploying directly on Render's Python web service runtime with zero manual configuration.
Free-Tier Design: Built specifically for Render's plan: free tier with zero mandatory paid dependencies or database requirements.
Production Server: Uses gunicorn with 2 worker processes binding dynamically to 0.0.0.0:$PORT.
Automated Health Checks: Monitored via /api/health for zero-downtime deploys.
Safe Ephemeral Cleanup: Uploaded archives are processed in ephemeral sandboxes and immediately released to respect Render container memory limits.

WildLens 🌿: An Open Field Naturalist Guide & Quest Engine to Get You Outside

Aayush Gupta — Sun, 11 Oct 2026 19:50:34 +0000

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass

What I Built

Most nature apps try to keep you looking at your phone. WildLens was built with the opposite philosophy: make the screen the shortest part of your outdoor experience.

WildLens is an open-source field companion and naturalist guide designed for hikers, curious neighborhood walkers, biology students, and families exploring local parks.

It does three core things:

Snap & Identify: Point your camera at any plant, wildflower, mushroom, bird, or insect. In seconds, you get taxonomic classification (common & scientific names, family), key distinguishing physical markers, and critical field safety precautions (toxic lookalikes, edible traits, safe handling).
Narrated Audio Field Guide: Instead of squinting at a bright screen on the trail, you can tap a button and listen to natural voice narration detailing species behavior and safety notes while keeping your eyes on the trail.
Seasonal Outdoor Quests: Select your environment (city park, conifer forest, wetland, coastline, or your own backyard) and WildLens builds a personalized, season-accurate nature scavenger hunt with interactive checklists and field observation hints.
Offline Field Journal: Log your sightings directly into an offline-first field journal with photos, notes, dates, and taxonomy tags—stored entirely on your device with zero remote tracking.

Demo

Live Application: https://wildlens-qkuh.onrender.com/
Curated Samples: If you're testing from your desk, the Identify page includes 1-click verified field samples (Monarch Butterfly, Fly Agaric, and Barn Owl) so you can test identification and audio narration immediately.

Code

The complete source code is public and open-source under the MIT license on GitHub:

ayushg-1506 / WildLens

WildLens 🌿

An intuitive field naturalist guide and biodiversity companion. Discover, identify, and document the living world around you.

Overview • Screenshots • Key Features • Tech Stack • Project Structure • Design Principles • License

Overview

WildLens is an open-source nature companion built to get people off their screens and back into the soil. Whether you're hiking a mountain ridge, strolling through a city park, or exploring your own backyard, WildLens transforms curious outdoor walks into rich ecological learning experiences.

Point your camera or upload a photo of any flower, tree bark, raptor, pollinator, or fungi. WildLens instantly delivers taxonomic classification, field safety warnings (toxic lookalikes, edible traits, handling precautions), distinguishing morphological markers, and narrated audio guides.

When you're ready to head outside, generate a Seasonal Outdoor Quest with habitat-specific scavenger hunts and log every observation directly to your private, offline-first Nature Field Journal.

🔗 Live Deployment…

View on GitHub

Direct link: https://github.com/ayushg-1506/WildLens

How I Built It

WildLens is built with a lightweight, modern open-source stack designed for responsiveness and privacy:

Next.js & Turbopack (App Router): Fast server and client architecture with clean, minimal bundling.
Multimodal AI & Taxonomy Engine: Uses Google's visual intelligence engine with a resilient multi-model fallback wrapper (src/lib/ai.ts) that extracts strict, structured JSON representing taxonomic hierarchy, lookalike distinguishing markers, habitat patterns, and safety warnings.
Natural Voice Narration with ElevenLabs: Powered by ElevenLabs for documentary-style naturalist field audio narration, with graceful fallback to browser speech synthesis when offline.
Geocoding via OpenStreetMap: Reverse-geocodes coordinates using OpenStreetMap Nominatim to calibrate quest items to local ecosystems without proprietary map SDK lock-ins.
Local-First Persistence: The nature journal runs entirely on browser localStorage, ensuring field notes, sightings, and photos stay private and accessible without cell service.
Organic Design System: Built with Vanilla CSS design tokens and Tailwind CSS using natural earth tones (forest deep greens, warm ochre, unbleached linen) and botanical typography (Playfair Display & DM Sans).

Why Does Open Innovation Matter?

When you are out in nature, software shouldn't get in your way:

Privacy by Default: Outdoor exploration often involves locations near homes, schools, and private trails. By leveraging local browser storage and stateless API inference, WildLens doesn't track users, build behavioral ad profiles, or require account logins.
Community Grounding: Nature is inherently open. Open software enables naturalists, park educators, and hikers to inspect how species safety warnings are structured, modify local flora data, and customize scavenger quests for regional parks.
Resilience & Independence: An open-source model allows swapping runtimes, adding local on-device small models, or hosting independently without fearing vendor lock-in or subscription paywalls on fundamental biodiversity knowledge.

Prize Categories

Best Use of Render: WildLens is hosted, built, and continuously deployed as a production Web Service on Render using a native render.yaml blueprint configuration.
Best Use of ElevenLabs: Features voice synthesis via the ElevenLabs API to deliver hands-free audio field guides on the trail.

StudySpark ⚡ - Turning Messy Notes into Smart Study Tools

Aayush Gupta — Sun, 04 Oct 2026 16:56:34 +0000

This is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend

What I Built

I built StudySpark for my college roommate. He always ends up with massive, chaotic walls of text for his lecture notes and spends way too much time trying to manually organize them before exams rather than actually studying them.

StudySpark solves this by instantly turning raw study materials into working study tools. You just paste your messy notes in, and it automatically generates a deck of spaced-repetition flashcards, adaptive quizzes, and clean summaries. If you get stuck on a specific concept, it even breaks it down for you using simple analogies.

Demo

Try it out live here: https://studyspark-njsw.onrender.com/

Code

ayushg-1506 / studyspark

⚡ StudySpark

Transform your chaotic lecture notes into working study materials instantly.

StudySpark is a sleek, AI-powered study assistant built for students who struggle to organize their textbook excerpts and lecture notes. By leveraging Google's Gemma open-weight models (via the Gemini API), it processes your notes to ensure private, secure, and fast learning.

✨ Features

Smart Flashcards: Automatically extracts key concepts from your notes and converts them into an interactive flashcard deck with spaced repetition marking.
Adaptive Quizzes: Generates multiple-choice quizzes to test your comprehension on the fly.
Key Summaries: Condenses walls of text into bulleted highlights and extracts key terms for quick review.
Concept Explainer: Stuck on a specific topic? Type it in, and StudySpark will break it down with analogies, real-world examples, and study tips.

🚀 Built With

Backend: Python, Flask, Gunicorn
Frontend: Vanilla JS, Custom CSS (Minimalist Design System)
AI…

View on GitHub

How I Built It

I wanted to keep the application lightweight and snappy, so I built a Python Flask backend with a completely custom Vanilla JS frontend (no heavy frameworks).

The heavy lifting is powered by Google's Gemma open-weight models (accessed via the Gemini API). I actually had to build a custom fallback system in the backend routing: because the larger gemma-4-31b model can sometimes hang during peak loads, the app dynamically falls back to faster models to ensure the flashcards generate instantly without crashing the student's study session.

Why Does Open Innovation Matter?

Study notes are deeply personal data. Students dump their half-baked thoughts, mistakes, and sometimes personal information into their lecture notes. Sending all of this into a closed, proprietary black-box API where you don't know how the data is used for training isn't ideal.

By building around open-weight models like Gemma, we are paving the way for tools like StudySpark to eventually run 100% locally on a student's laptop. That means total privacy, offline capability in library basements, and most importantly for broke college students: zero recurring API costs.

Prize Categories

Best Use of Gemma
Best Use of Render