Opensecagent

A next-generation platform for building, deploying, and managing autonomous AI agents — with a focus on security, automation, and simplicity.

Comment

Inspiration

We were inspired by two converging trends: the rise of autonomous AI agents and the growing complexity of securely deploying them. While LLMs are transforming how we build tools, deploying multi-agent systems often requires deep DevOps knowledge and complex cloud setup. We wanted to make it as easy as describing your agent in plain English — and let the system handle the rest. Opensecagent was born to enable this seamless, secure, and intelligent deployment path.

What it does

Opensecagent turns a user prompt into a fully functional, cloud-hosted AI agent. It auto-generates agent code, builds a GitLab CI pipeline, configures Docker and Cloud Run deployment, and hosts live API endpoints. It also supports a registry of agent templates, secure DevSecOps defaults, and multi-agent collaboration features. In short, it's an end-to-end AgentOps platform that helps anyone go from idea to deployed AI in minutes.

How we built it

We built a frontend wizard using React + TypeScript that dynamically adapts based on user inputs and intelligently suggests agent types. The backend uses Netlify Functions to process the configuration, generate code artifacts (e.g., agent.py, .gitlab-ci.yml, Dockerfile), and interact with the GitLab API to create repositories and push code. Google Cloud Build is used to create Docker images, Artifact Registry stores them, and Cloud Run serves them securely and scalably.

Challenges we ran into

  • Navigating Google Cloud IAM and permission errors during service account impersonation
  • Managing CI builds across GitLab runners with limited default tools (e.g., missing bash, curl)
  • Handling edge cases in prompt-to-code generation for diverse agent types
  • Ensuring end-to-end flow reliability: from prompt input to Cloud Run service health validation

Accomplishments that we're proud of

  • Successfully implemented prompt-to-agent pipeline with live deployment
  • Built a functional multi-agent deployment framework within a short hackathon timeframe
  • Generated live Cloud Run endpoints that support /process, /health, and dynamic configs
  • Created reusable and customizable templates (e.g., Curie, Atlas, Deng) for future scalability

What we learned

  • How to integrate CI/CD pipelines tightly with frontend user flows using GitLab APIs
  • The importance of pre-validating GCP permissions and providing clear logs in CI stages
  • Prompt-based UX can significantly reduce AI deployment friction — even for non-engineers
  • AgentOps is not just about deployment, but lifecycle control, security, and collaboration

What's next for Opensecagent

  • Expand prompt-to-agent capabilities with LangChain and HuggingFace integrations
  • Add persistent vector storage and multi-agent memory coordination
  • Introduce team-based agent dashboards, role-based access, and audit trails
  • Support agent-to-agent messaging and external triggers (e.g., Slack, Webhooks)
  • Launch Opensecagent Registry: a marketplace for open-source and commercial agent templates

Built With

Share this project:

Updates