See our Notion Page for the resources guide.

Welcome to Hack for Impact NYC 2025! This resource guide is here to help you along your hackathon journey – especially if you’re a beginner. Hackathons can sound intimidating, but they’re actually fun, friendly events where you get to learn and build something cool in a short time. You don’t need to be a coding pro or a tech genius to take part – just bring your curiosity and enthusiasm. We’ve gathered some tips and links to tools, inspiration, and support to get you started.

This year’s hackathon focuses on four impact tracks: Education Equity, Climate Resilience, Urban Challenges, and Responsible AI. The resources below are general-purpose and beginner-friendly – no heavy jargon, we promise – so you can use them no matter which track you’re tackling. Dive in, and remember: the goal is to learn, collaborate, and make an impact. You’ve got this!

• MCP-Agent: is an open-source framework designed to streamline the creation of production-ready agents by using MCP server and composable agent patterns [Guide].
• LangChain: Widely adopted for building language model-based agents with memory, tool use, and workflow support
• AutoGen: Open-source framework for creating collaborative, multi-agent systems and orchestrating agent workflows
• CrewAI: Python-based toolkit for building role-based, multi-agent teams that coordinate on complex tasks
• SuperAGI: Open-source platform for fully autonomous agents that manage and execute tasks independently
• AgentGPT: User-friendly interface for deploying autonomous agents, ideal for quick prototyping and smaller projects
• Botpress: Visual, no-code/low-code platform for building conversational AI agents with multi-channel support
• LlamaIndex: Framework focused on data management and retrieval for agents working with large datasets

• Lovable – AI turns a napkin idea into a working full-stack web app in minutes; great when you have no devs on the team.
• Bubble – Visual drag-and-drop builder that now embeds GPT-style agents to scaffold back-end logic; huge template library.
• Softr – Easiest on-ramp: connect Airtable/Google Sheets and publish a secure client portal or internal tool fast.
• Glide – Converts spreadsheets into mobile/desktop apps; new AI column type lets novices add chat or image-gen features.
• FlutterFlow – Visual builder that exports clean Flutter code; AI assist drafts whole screens from a prompt.

• Cursor – AI-first VS Code fork that predicts edits and answers questions about your codebase; just hit Tab × 3.
• Windsurf (Codeium) – Free autocomplete + chat inside every major IDE; enterprise "Windsurf" editor adds agent-style refactors.
• Tabnine – Privacy-focused completion engine now shipping a Dev Preview and code-review agent for teams.
• GitHub Copilot – The original "AI pair programmer", recently upgraded with Autofix and forthcoming rate-limited tiers.

• Aider – Chat with GPT-4 in your shell to add files, run tests, and commit changes; perfect for CLIs or headless servers.

• Scribe – Click-through recorder that auto-generates step-by-step guides for any workflow (great for demo hand-offs).
• Mintlify – Beautiful developer docs site plus an AI chat sidebar; VS Code Doc Writer extension drafts JSDoc for you.
• Swimm – Keeps docs "code-coupled"; AI watch-dog updates snippets when the code changes, so tutorials never rot.
• DocsGPT – Open-source assistant that plugs ChatGPT into your docs repo—handy if you want self-hosted Q&A.
• CodeX Docs – Lightweight, self-hosted Docs-as-Markdown system built on Editor.js; easy for small teams.
• Apidog – One-stop API tester that also autogenerates OpenAPI specs and interactive docs from your endpoints.

• ParentLine - SMS auto-translator
  • What it does: Forward-and-back translation between teachers and immigrant parents
  • Stack in a flash: Twilio SMS → Google Translate API → Supabase table
  • Why it's proven: TalkingPoints scales this exact model to 150 languages today (talkingpts.org), showing the workflow is production-ready and easy to replicate at hackathon scale
• Clip-to-Class - lesson-recap bot
  • What it does: Teachers drop a Zoom/Meet recording link; the bot returns a timestamped summary plus vocabulary list for English-learners
  • Stack: OpenAI Whisper for transcription → GPT for summarisation → simple React front-end (can be scaffolded with Vite) → GitHub deploy
  • Feasibility evidence: Weekend hackers used the same OpenAI + Next.js combo to win an AI hackathon in 48 h (DEV Community)

• WasteSortCam - smart recycling kiosk
  • What it does: Webcam or phone cam tells you which bin to use
  • Stack: TensorFlow-Lite model retrained on TrashNet images → runs locally in a PWA
  • Proof it fits the clock: "TrashCam" was fully coded and demoed in a 24-h student hack and is open-sourced on Devpost (Devpost - The home for hackathons); a similar mobile build ("Recycling Sorter") hit production accuracy with transfer-learning in one weekend (Devpost - The home for hackathons)
• Footprint 411 - personal CO₂ receipt
  • What it does: User snaps a food or product barcode; app returns estimated embodied carbon and a greener alternative
  • Stack: OpenFoodFacts or USDA API → public LCA lookup sheet → small Svelte front-end
  • Why doable: Smart-city hackers already piped real-time IoT sensor data into dashboards in 24 h (Technical.ly); swapping sensors for a ready-to-query database shortens the path further

• RatMap Mini
  • What it does: Drop an address, see latest rat-inspection results and auto-generated "next steps" email template to your landlord
  • Stack: NYC Rat-Sighting JSON feed (NYC Open Data) → Leaflet map → GPT prompt for email text
  • Precedent: Open-data gallery highlights dozens of similar civic mash-ups launched in single-day jams (opendata.cityofnewyork.us)
• NoiseSniffer Lite
  • What it does: Turns any spare Android phone into a decibel logger; uploads every 5 min to a shared map so neighbourhoods see real noise hot-spots
  • Stack: Expo React-Native starter → phone mic API → Supabase row insert
  • Inspiration: The full Noise Sniffer concept clinched second place at a 24-h smart-city hackathon (Technical.ly)—your cut-down MVP needs only software

• BiasCheck Bot
  • What it does: Drag-and-drop CSV → bot returns fairness metrics and mitigation suggestions
  • Stack: IBM AI Fairness 360 Python library (pip install) exposed via a FastAPI endpoint → simple React UI
  • Evidence: AIF360 ships with example notebooks demonstrating results in minutes (IBM Research, GitHub), and student hackathons regularly use it to tackle bias detection challenges (Responsible AI Hackathon)
• Model Card Maker
  • What it does: Query a deployed Hugging Face model, auto-compiles a model-card draft (datasheet + risks) ready for GitHub Pages
  • Stack: Hugging Face Inference API → templated Markdown → Netlify one-click deploy
  • Precedent: Fairness-focused hackathons emphasise lightweight documentation tools as a first deliverable, showcased in bias-assessment events across 2024-25 (The Dialogue)

You’re never on your own at Hack for Impact NYC. Use these channels to get answers fast: