Build, experiment, and ship working AI-agent projects with developers from across Philadelphia.

Join developers from across Philadelphia for a hands-on hackathon focused on building with AI agents.

Expect practical challenges, working prototypes, and experimentation with agentic systems. You can arrive with a team, meet collaborators at the event, build independently, or join without a finished project idea.

Developers, engineers, technical builders, and anyone interested in experimenting with AI-agent systems.

Whether you're experienced with agent frameworks or just starting to explore them, the goal is to build, test ideas, learn from others, and leave with something working.

Two days. Two locations. One hackathon.

Challenge tracks and awards are being finalized with participating partners.

The Quirq Challenge: build an agent—or try to break its workspace.

Agents can run for hours, consume tokens, modify files, and call tools—but it is often difficult to see what they actually did or determine whether they stayed within their intended boundaries. That is the problem Quirq works on.

This challenge has two tracks: Build It and Break It.

Quirq: Build It

Build any useful agentic system and run it through Quirq, using either the managed cloud at app.xo.builders or the local runtime. You may use Claude Code, Codex, OpenClaw, Hermes, n8n, your own agent, or another runtime.

Your project can solve any problem and may also compete in another hackathon track. What matters here is that your agent performs meaningful work and that its activity is visible through the xo-space: its environment, sessions, actions, files changed, costs, and results.

You are not only being judged on whether your agent works. You are also being judged on whether people can understand what it did, evaluate its output, and trust how it operated.

Judging criteria:

• Agentic Design & Execution — 30%
• Observability & Space Display — 25%
• Impact & Usefulness — 25%
• Reliability & Safety — 20%

Prize: $500 cash plus an equivalent amount in Quirq credits. One winner.

Quirq: Break It

Test the security and integrity of your own cloud-managed Quirq workspace at app.xo.builders.

Look for ways to escape the intended workspace scope, conceal agent activity from the xo-space, misreport usage, bypass controls, or otherwise cause the workspace to behave in an unsafe or unexpected way.

Each finding must be submitted privately and include clear reproduction steps, an explanation of its impact, and a suggested fix. Do not test other participants’ workspaces, disrupt shared infrastructure, or target Code & Coffee systems, the local Quirq installation, or third-party model providers.

Validated findings earn points based on:

• Severity and impact
• Reproducibility
• Novelty
• Scope and systemic impact
• Quality of the report

Bounty: Up to $500 total may be distributed across accepted findings. Awards depend on the quality and severity of validated findings; the full bounty is not guaranteed. Invalid, duplicate, low-impact, or non-reproducible reports may receive reduced points or no payout.

THE CODE REGISTRY CHALLENGE: BUILD FAST, THEN PROVE IT HOLDS UP.

Today you will generate a lot of code, and most of it will be written by an agent. Almost none of it will be read by a human. That is the problem we work on.

The Code Registry analyses a codebase and returns a Code Score out of 1,000 across three pillars: Security, Dependencies and Quality. There is no LLM anywhere in the analysis pipeline. It is deterministic, so the same code returns the same score every time.

This track runs across the whole event. Build whatever you want, in any language, for any other track. Then run your project through The Code Registry before the cutoff. The highest final Code Score wins.

You are not being judged on whether your agent works. You are being judged on what your agent produced.

Prize: a 12-month Code Registry plan covering up to 400,000 lines of code, listed value $3,600. One winner.

HUMANSTANDARD TRACK: BUILD FOR HUMAN MUSIC

Description:

HumanStandard is a forensic AI music detection company. Our API takes a track and returns a signed verdict on whether the audio was generated by AI, performed by humans, or is a hybrid of the two, along with the evidence behind that call. Distributors, DSPs, and rights holders use it to comply with the EU AI Act, DDEX labeling standards, and platform policies on AI content.

For this track, build anything that uses the HumanStandard detection API to solve a real problem in music. Some directions to get you started: a submission screener for a label or playlist curator, a browser extension that flags AI tracks on streaming pages, an artist tool that generates proof of human authorship, a catalog auditing dashboard, a Discord bot for music communities, or an agent that investigates suspicious releases end to end. We care most about ideas that a real person in the music industry would actually want to use.

This track is built for people who want to go into music tech. Beyond credits and a fellowship, the winning team gets direct introductions from Rasha Rahman, HumanStandard's founder, to companies across the music industry, and a working session on turning the project into something real. If you are looking for a way into this world, this is the track for you.

What you get during the hackathon:

• A HumanStandard API key issued to your team on the morning of the hackathon, preloaded with 200 credits. One credit is one song scan. Keys are generated from the team email you provide at registration, so make sure that email is one your team actually checks.

• API documentation and a quickstart you can hand to a coding agent.

• Arian from HumanStandard will be on site all day to help with the API and answer questions. Rasha Rahman, HumanStandard's founder, will also be available throughout the hackathon and will hold an office hour window where you can bring your project for a critique.

• Bring Your Own Project - Technical Execution
  How well does the project work? Is the implementation technically sound and functional?
• Bring Your Own Project - Agentic Design
  How meaningfully does the project use agents, tool calling, memory, planning, multi-agent workflows, or autonomous execution?
• Bring Your Own Project - Innovation & Creativity
  How original is the idea, approach, or use of agent technology?
• Bring Your Own Project - Impact & Usefulness
  Does the project solve a meaningful problem or provide clear value to users?
• Bring Your Own Project - Reliability & Safety
  Does the team account for failure modes, evaluation, permissions, guardrails, or other reliability and safety considerations?
• Bring Your Own Project - Demo & Completeness
  How clearly does the team demonstrate what they built, and how complete is the working prototype?
• Human Standard - Does it work?
  The project runs and does what the team claims. The HumanStandard API is actually integrated and the verdict data is used in a meaningful way, not just displayed.
• Human Standard - Is it a real problem?
  The team identified a genuine pain point for someone in the music industry, whether an artist, label, distributor, curator, platform, or fan. Bonus points for teams that can name who would use this and why they would care.
• Human Standard - Product quality
  The user experience is thoughtful. Verdicts are communicated honestly and with appropriate uncertainty rather than as a binary "this is AI" claim. The project feels like the start of something a person could ship, not just a proof that the API responds.
• Human Standard - Music tech ambition
  The team shows real curiosity about the music industry and a desire to keep building in it. This matters because the prize is largely career oriented: introductions, mentorship, and a fellowship are most valuable to people who want to pursue this path.
• Quirq Build it - 30% weight; Agentic Design & Execution
  Meaningful use of agents, including tool calls, memory, planning, multi-agent collaboration, and autonomy. Most importantly: does the system actually work?
• Quirq Build it - 25% Observability & Space Display
  How clearly the demo shows the environment, agents, and their activity—including files touched, sessions, and costs. The Quirq space must be part of the demo.
• Quirq Build it - 25% Impact & Usefulness
  Whether the project solves a real problem for a real user.
• Quirq Build it - 20% Reliability & Safety
  How well the project handles failure modes, scoped permissions, guardrails, and evaluation.
• Quirq Break It Bounty
  Quirq Red Team Bounty is evaluated separately using a points-based system: Severity and impact, Reproducibility, Novelty, Scope or systemic impact, Quality of the report
• The Code Registry Bounty
  Eligible projects will be analyzed using The Code Registry. The participant whose hackathon code receives the highest final Code Score will win.