AI Hackathon
← DEPARTURES
AI AgentsBoarding

Delphi · Agent Arena Competition: Evidence Mining

Delphi, Gensyn · on Dorahacks

Cash prize: $7,250
Free credits: Credits: not stated
Who can enter: Solo OK Country rules: unknown
Code rights: Not checked yet
DelphiGensyn
Registration closesDec 11

Build software that gathers and weighs evidence to improve forecasts in blockchain-based markets.

Proceed to gate Opens dorahacks.io. You register on the organizer's site.

Imported from Dorahacks and approved by our team. We haven't checked the rules by hand. Confirm the rules on the event page before you enter.

One email a week

New events that match your topics, every Monday. Unsubscribe anytime.

Flight details

Registration closes:
2026-12-11 17:00 UTC
Submissions close:
2026-12-11 17:00 UTC
Event ends:
2026-12-11 17:00 UTC
Cash prize:
$7,250

Prizes listed: Total cash prizes: USD 7,250 (paid in USDC; 1st $3,000, 2nd $2,000, 3rd $1,000, 5x $250 feedback bounties)

In short

  • Mining opens 23 November 2026 and closes 11 December 2026 at 17:00 UTC on DoraHacks.
  • The total prize pool is USD 7,250 in USDC across leaderboard places and feedback bounties.
  • Solo entrants and teams are both welcome, with at least 10 judged submissions across 3 markets required.

Project ideas

Delphi Evidence Graph

An agent gathers claims from public sources and scores each one so forecasters can see which evidence is strong.

Architecture: Ingest market questions and source links. → Extract claims and link them to sources. → Score each claim for recency, independence and clarity. → Publish a cited evidence brief for forecasters.

Stack: Python, LangGraph, Gensyn, Delphi

Manipulation Watch Agent

An agent flags coordinated claims and weak sources before they distort a forecasting market.

Architecture: Collect market questions and source posts. → Cluster repeated claims and shared accounts. → Detect anomalies and rank source risk. → Generate a review report with citations.

Stack: Python, scikit-learn, Gensyn, Delphi

Cited Evidence Swarm

A swarm of agents splits source research, cross-checks findings and returns a cited evidence packet.

Architecture: Split each market question into source tasks. → Assign tasks to parallel research agents. → Merge findings and remove duplicate citations. → Score confidence and return the evidence packet.

Stack: Python, AutoGen, Gensyn, Delphi

FAQ

Can I enter on my own?
Yes. The official DoraHacks page states that solo entrants and teams are both welcome, using one registered wallet per entry.
Who is eligible to take part?
Builders with research agents or scrapers can join. Prize eligibility requires integrity review, sanctions screening and KYC under the official Terms of Service.

Sources

Open the event page →

More open AI hackathons