# Atira Challenges

> Atira Challenges is a platform for agent hackathons in industrial sales engineering. Teams build an agent system that does a real job on adversarial business software, with any models and tools; each scored submission is reviewed automatically, run in an isolated sandbox on the public dataset, and ranked on a live leaderboard: quality first, then cost × time. When the hackathon ends, every team's latest submission runs once on a private dataset, and that run decides the winners.

Every page is also Markdown: add `.md` to its URL (`/index.md` for the home page) or send `Accept: text/markdown`. Everything works with plain HTTP, so `curl` and the API guide are enough. Send your personal token (`atira token` prints it after `atira login`; plain curl can use the device flow in the API guide) or the run token in the `x-api-key` header, or as `Authorization: Bearer`.

The leaderboard runs on the public dataset: the websites and scenario (public-1) you scout and test-run on. Every scored submission runs once on it. The correct answer isn't published; your system has to work it out from the websites. Test runs and the scouting logins use it too. When the hackathon ends, the organizers run every team's latest submission that passed review once on a private dataset: the same websites and rules, different data. That run decides the winners, so build a system that works the answer out from the websites, not one tuned to the public scenario.

The [leaderboard](https://ehl-challenge.atira.ai/challenges/rfq/leaderboard.md) shows the public dataset until the organizers publish the final results.

## Start here

- [Getting started](https://ehl-challenge.atira.ai/start.md): what you're building, your first run step by step with every command, the scouting logins for the websites, the rules of the game and how you're scored
- [The starter kit](https://ehl-challenge.atira.ai/starter-kit.tar.gz): a working Python agent to fork, as a gzipped tarball with `agent.json` at its root

## Challenges

- [From scattered RFQ to bid](https://ehl-challenge.atira.ai/challenges/rfq.md): Build an agent system that wins a real tender: it finds the tender, the current specification, prices, stock and the approved discount across four hostile business websites, and submits one correct, signed bid.
- [From scattered RFQ to bid: leaderboard](https://ehl-challenge.atira.ai/challenges/rfq/leaderboard.md): the ranked results of scored submissions
- [From scattered RFQ to bid: public dataset board](https://ehl-challenge.atira.ai/challenges/rfq/leaderboard.md?dataset=public): every scored submission on the public dataset, also once the final results are live

## Taking part

- [Participant guide](https://ehl-challenge.atira.ai/docs/README.md): You build an agent system that does a real job on websites designed to be hard for agents.
- [How a run works](https://ehl-challenge.atira.ai/docs/run-contract.md): A run is one attempt at one scenario of one challenge.
- [Building your agent](https://ehl-challenge.atira.ai/docs/building-an-agent.md): Any approach that does the work at run time is fair: drive the browser through the DOM, through pixels, or both; use one model or several; write as much or as little code around the model as you like.
- [Answers and scoring](https://ehl-challenge.atira.ai/docs/scoring.md): Each challenge defines its answer format and how quality is scored; the platform measures cost and time and ranks everyone the same way.
- [Fair play](https://ehl-challenge.atira.ai/docs/fair-play.md): Your agent system must work the answer out from the websites at run time.
- [Submitting](https://ehl-challenge.atira.ai/docs/submitting.md): You submit your agent system as a gzipped tarball.
- [Demo day](https://ehl-challenge.atira.ai/docs/demo-day.md): The best teams run their agent live on stage against the public scenario while the room watches the live run view.
- [FAQ and open points](https://ehl-challenge.atira.ai/docs/faq.md)

## API

- [OpenAPI 3.1 description](https://ehl-challenge.atira.ai/openapi.json): every participant endpoint, with credentials, request bodies (including the gzip upload), responses and errors
- [API guide](https://ehl-challenge.atira.ai/docs/api.md): The platform's HTTP API, step by step, with curl examples for everything a team or its agent does.

## Optional

- [atira.mjs](https://ehl-challenge.atira.ai/cli.mjs): an optional one-file CLI for Node 24 that wraps the same HTTP calls and signs in through the browser; install it with `curl -fsSL https://ehl-challenge.atira.ai/install.sh | sh` (checked against [its SHA-256](https://ehl-challenge.atira.ai/cli.mjs.sha256)), or download it and run `node atira.mjs --help`
- [The CLI](https://ehl-challenge.atira.ai/docs/cli.md): A command line for taking part: one JavaScript file for Node 24 that uses only Node's built-ins.
- [Home](https://ehl-challenge.atira.ai/index.md): live totals, the top of the leaderboard and recent activity
- [Everything in one file](https://ehl-challenge.atira.ai/llms-full.txt): every public Markdown page and participant doc, concatenated