Build the Future of Work with AI Agents
Sitrep turns meetings into tasks. Build an AI agent that automates one of them — then publish it to the Sitrep Marketplace, where every valid submission goes live for businesses to install and use.
The challenge
Turn a meeting task into finished work
After every meeting, Sitrep produces a summary and a set of action-item tasks. It hands one of those tasks to your agent — and your agent returns a useful artifact. Unlike a typical hackathon, every valid submission gets published to the Sitrep Marketplace for businesses to install and use.
1. SitRep captures the meeting
The notetaker bot joins, transcribes, and turns the conversation into a summary plus action items.
2. A task reaches your agent
SitRep sends your endpoint a signed request with the task, meeting summary, and attendees.
3. Your agent ships an artifact
You return a markdown draft, an HTML document, or a link — whatever makes the task done.
Artifacts your agent could produce:
Two ways to compete
Pick your track
Both tracks are hosted the same way — run locally with a tunnel, or deploy to a free host.
No-code
Edit prompt.txt
Pure prompt engineering. Perfect for fast ideas and iterating on what a great assistant should write — no Python required.
Code
Edit handler.py
Full control: tools, multi-step chains, file generation, and external APIs. Start from a working example handler and make it yours.
How it works
From fork to Marketplace in four steps
Create a free Sitrep account
Sign up at joinsitrep.com — no card required. Everything you build lives there.
Pick a track & build
No-code: configure a system prompt, model, temperature, and triggers. Code: use the Starter Kit to call APIs, databases, and MCP servers.
Publish to the Marketplace
Ship your agent to the Sitrep Marketplace so businesses can install and use it — every valid submission goes live.
Submit on Kaggle
Add your published agent URL and a public MIT-licensed repo to your Kaggle writeup before the deadline (and star our repo).
You bring the LLM — a free local model via Ollama, or your own key. SitRep never charges you for tokens and never runs your code; it just calls your endpoint.
The rules
What a valid submission needs
- Publish your agent to the Sitrep Marketplace and include its link.
- Write a Kaggle writeup (≤1000 words) covering what you built and how.
- Make your code public in a GitHub repo with an MIT license.
- Star the SitrepNoteTaker repo — it's a required step.
Timeline
Hackathon opens
Jul 13, 2026
Sign up free at joinsitrep.com
Submissions close
Jul 20, 2026
Publish your agent + submit your Kaggle writeup
Marketplace Choice Award
~30 days later
Most installs, active usage & ratings
Prizes
- No-Code Track$450
Winner $250 · Runner-up $125 · Third $75
- Code Track$450
Winner $250 · Runner-up $125 · Third $75
- Marketplace Choice$100
Awarded 30 days later by real usage
Judging
Business Impact
Solves a real problem — 30 pts
Agent Quality
Accurate, reliable output — 25 pts
User Experience
Clear, polished, intuitive — 15 pts
Innovation
A creative workflow or use of AI — 15 pts
Execution
Code architecture or prompt design — 10 pts
Docs & Demo
Clear writeup and demo — 5 pts
Under the hood
A lean starter, up in ~10 minutes
A small FastAPI service with the SitRep request contract already wired. No heavyweight agent framework — just Python, httpx, and your ideas.
Quickstart
# 1. Get a free local LLM (or bring your own key)
ollama pull llama3.1
# 2. Configure + run
cp .env.example .env
bash scripts/run-local.sh # http://localhost:9000
# 3. Smoke-test (new terminal)
bash scripts/smoke-test.shThe contract
// SitRep POSTs to <your-url>/run
{ "task": {...}, "summary": "...",
"attendees": [...], "agent": {...} }
// You respond with artifacts
{ "artifacts": [
{ "type": "markdown" | "html" | "link",
"title": "...", "content": "..." } ] }FAQ
Good to know
Do I need to pay for an LLM?
No. You bring the LLM — a free local model via Ollama works out of the box, or bring your own key (OpenAI, OpenRouter, vLLM, LM Studio). SitRep never charges you for tokens.
Does SitRep run my code?
Never. SitRep only calls your HTTPS endpoint with a signed request. Your agent runs wherever you host it.
No-code or code — which track should I pick?
No-code (prompt.txt) is perfect for prompt-engineering ideas and fast iteration. Code (handler.py) unlocks tools, multi-step chains, file generation, and external APIs. Both are hosted the same way.
What if I can't self-host at all?
You can build a Managed (no-code) agent entirely in the SitRep Studio — prompt only, SitRep runs it, no repo or server needed. The starter kit is for Remote agents you host yourself.
Love open source? Star SitrepNoteTaker
Our open-source meeting notetaker — and starring it is part of a valid Kaggle submission.
Ready to build the future of work?
Grab the starter kit, ship an agent, and get it in front of the SitRep community.
