Build an agentic application using CockroachDB for persistent memory and deploy it on AWS to compete for $8,750 in cash prizes.
By Cockroach Labs
The CockroachDB × AWS Hackathon challenges developers, engineers, and AI builders to create agentic applications that leverage CockroachDB's distributed AI capabilities and AWS services. The hackathon focuses on building AI agents with production-grade, persistent memory, recognizing that reliable memory is crucial for agents transitioning from experimental stages to real-world production workflows. AI agents require memory that is always available, distributed, and scalable, unlike traditional databases optimized for human-scale interactions. CockroachDB is presented as the ideal system of record for agentic memory, offering global distribution, continuous availability, PostgreSQL compatibility, and native integration into the agent toolchain through its Managed MCP Server, ccloud CLI, and an open-source Agent Skills ecosystem. This robust foundation ensures memory persists across regions, failures, and scales without data loss or maintenance. Participants are invited to innovate using CockroachDB's tools, including the Managed MCP Server, Distributed Vector Indexing, agent-ready ccloud CLI, and the open-source Agent Skills Repo, all deployed on AWS. The goal is to demonstrate how a production-grade memory layer makes an agent useful and resilient in real-world scenarios, addressing the critical need for agents that can reliably remember, think, and act.
Build an agentic application that uses CockroachDB as its persistent memory layer, deployed on AWS. Your agent should store, retrieve, and act on memory, which can include conversation history, user context, task state, embeddings, or structured transactional data. The submission must demonstrate that memory is integral to the agent's production utility. All submissions must use at least two of the following CockroachDB tools: - CockroachDB Cloud Managed MCP Server - CockroachDB Distributed Vector Indexing - ccloud CLI (Agent-Ready) - CockroachDB Agent Skills Repo (Open Source) All submissions must also use at least one AWS service: - Amazon Bedrock (foundation models, knowledge bases, or agents) - AWS Lambda (serverless agent execution) - Amazon ECS / EKS (containerized agent workloads) - Amazon S3 (artifact or document storage) - Amazon SageMaker (model training or inference) - Amazon Bedrock Agents (multi-step agentic workflows) - Any other AWS service that powers your agent's environment
The hackathon is open to individuals who are at least 18 years old (or have reached the age of majority in their jurisdiction of residence) and teams of up to 5 individuals, as well as organizations (corporations, non-profits, LLCs, partnerships). An eligible individual may join multiple teams/organizations or enter individually. The hackathon is not open to residents of or organizations domiciled in countries where participation or prize receipt is prohibited by U.S. or local law (e.g., Brazil, Quebec, Russia, Crimea, Cuba, Iran, North Korea, and OFAC-designated countries). Promotion Entities, their employees/agents, judges, and immediate family/household members, and any individual/organization whose participation would create a conflict of interest are also ineligible.
Answers below are best-effort estimates from the most recent application cycle. Values marked ~ or (est.) may shift year-to-year — confirm specifics on the official scholarship page.
No. CockroachDB Cloud has a free tier that is fully eligible for this hackathon. You can spin up a cluster in minutes at cockroachlabs.cloud with no credit card required.
Yes. The CockroachDB MCP Server supports any MCP-compatible client, and the Agent Skills Repo is model-agnostic; it works with Claude, Cursor, LangChain, or your own agent framework. You must use at least one AWS service, but you can use any combination of models.
Perfect. The hackathon starter kits are designed to get you from zero to a running agent in under 30 minutes. CockroachDB is fully PostgreSQL-compatible, so if you know Postgres, you already know most of CockroachDB.
Yes, and this is one of CockroachDB's key differentiators for agentic workloads. CockroachDB's integrated pgvector support with distributed indexing means you can store embeddings and transactional data in one system, eliminating ETL complexity and consistency issues between a separate vector store and your operational database.
MCP is an open standard created by Anthropic that allows AI agents to safely and predictably interact with external systems: tools, databases, APIs, through a structured, auditable interface. CockroachDB's Managed MCP Server implements this protocol to give AI coding agents like Claude Code and Cursor a direct, secure connection to your database cluster.
Agent Skills are structured, machine-executable capabilities published in CockroachDB's open-source skills repository. Each skill encodes a specific CockroachDB workflow like 'profile statement fingerprints' or 'detect schema anti-patterns' with clear inputs, outputs, and behavior. Skills follow open, standard interfaces so they work across models and agent frameworks without rewriting integrations.