HydraDB x Connectors Hackathon
HydraDB Connectors Hackathon
July 31, 6 PM PT → August 7, 6 PM PT · Virtual · $500 in prizes
Discord: Join Discord Server
HydraDB is a context layer for AI agents, combining knowledge, memories, documents, metadata, and relationships across multiple sources.
Challenge
Build a HydraDB project using at least 3 connectors and test how well it handles real cross-source retrieval.
Connectors
Twitter/X
Attio
Slack
Linear
Notion
Gmail
Bigtable
Dropbox
Google Drive
Confluence
Calendly
Google Calendar
Intercom
Jira
PostHog
Shortcut
GitHub
GitLab
Freshdesk
Stripe
use the free account on hydradb.com
What to Build
1. Multi-Connector + Document Ingestion
Ingest the same people, companies, projects, or events across multiple sources.
Example: the same person appears in Slack, Gmail, and uploaded documents. Test whether HydraDB connects those references instead of treating them as separate entities.
2. Difficult Retrieval Questions
Ask questions that require more than basic semantic search:
Temporal reasoning
Metadata filtering
Entity deduplication
Knowledge updates
Third-party attribution
Actor-based queries
Thread understanding
Multilingual retrieval
Multi-hop reasoning
Multi-hop questions are especially valuable. The answer should require multiple HydraDB queries or retrieval steps, not one API call.
Example:
Who filed
BUG-123, which project are they working on, and what did they say about the fix in Slack?
3. Latency vs Accuracy Competition
Design ingestion and retrieval that is:
Accurate
Fast
Cheap
Use:
Fast queries
Thinking queries
Metadata filters
The goal is to answer most questions using fast mode, while using thinking mode only when the question actually requires deeper reasoning.
Track:
Accuracy
Latency
Number of HydraDB calls
Fast vs thinking usage
Estimated retrieval cost
Submission
Include:
At least 3 working connectors
Document ingestion
Difficult cross-source questions
Expected vs actual answers
Latency and accuracy results
A 60-second demo
Judging
Correctness, cross-source reasoning, latency, cost, reproducibility, and developer experience.
How far can you push fast mode without sacrificing accuracy?
Resources
MCP: usecortex/hydradb-mcp
CLI: usecortex/hydradb-cli
Docs: https://docs.hydradb.com