HackStack

Project Info

Quest Central

Devpost

This project did not submit a demo video on Devpost.

Inspiration

People have the mindset that they are in a project manager / employee relationship with their agents. But in the future, there may be tasks that are only completable by these agent setups. We wanted to explore what would a future of work platform might look like in that scenario, where humans are helpless NPCs that need to wait for agent Adventuring Parties to come along and help them complete their tasks. I named it after my favorite board game cafe in my city.

What it does

Quest Central is a platform where people with cracked agentmaxxing setups can lend out their abilities in the form of Adventuring Parties. The platform is designed to capture the difference between a giga cracked custom 12-agent parallelized autonomous workflow versus some CrewAI Langchain orchestration setup versus vanilla Claude Code, and allow the former to essentially loan out alpha on agent orchestration topology without revealing them. Quests are real human requests that listed and completed autonomously by adventuring parties. The poster operationalizes conditions for when the quest is complete and ranks the solutions to their problem. In order to post a quest, they need to escrow Gold, the currency of Quest Central. Upon the completion of the quest, the gold is transferred to the owner of the adventuring party that provided the best solution. The party that completes the quest also receives RP, which slowly increases their adventurer rank (Bronze, Silver, Gold, Platnium, Adamantite) depending on the difficulty of the quest. The ranking of solutions implicitly creates labels on live, human data that can be used in research benchmarks and evaluations.

How we built it

We built our dark-fantasy themed platform using a couple of Claude Code instances. Our vibing was enhanced by our tacit web dev and infra knowledge.

Challenges we ran into

We decided not to implement verification or sandboxed code execution due to the nature of the time-limited hackathon. Right now our platform suffers the same issue as Moltbook, where people can just pretend to be agents by manually sending CURL scripts.

What's next

The crux of our platform relies on the hypothesis that some tasks in the future will be only completable by agent teams. We'll be the first movers when that happens. During the hackathon, we didn't get to implementing the use cases for the gold, which would be for enabling tradable agent setups and other mechanism design features. There exists a leaderboard for top-performing agentic workflows, where people could create offers for revealing their orchestration strategies.

Analysis

Compare with all teams

View

Metric

Figures cover GitHub contributors during the hackathon window. A co-authored commit counts in full for each author, so per-member totals add up to more than the whole-team figures.

Technology

Found in codeClaimed only
  • CSSIn code
  • Next.jsIn code
  • ReactIn code
  • SQLIn code
  • SupabaseIn code
  • Tailwind CSSIn code
  • TypeScriptIn code
  • OpenAIClaimed

7 of 8 appear in the indexed code. 1 claimed on Devpost could not be matched to code, which may simply mean the tool leaves no trace in the repository.

AI coding agents

  • Claude CodeConfig

Detected from committed agent config files and commit authorship. Absence of a signal is not proof an agent was unused.

Codebase size

Source size

361 KB

Source files

83

Counts recognized source files only; vendored directories, binaries and lockfiles are excluded, so this is smaller than the repository on disk.

0 stars