Access Your AI Development Team: VM Direct Access Explained
Your AI development team runs on a dedicated cloud VM — and now you can log in and work alongside them.
Synthcore's VM direct access gives you SSH access to the same environment where your 13 AI agents work, commit, and ship code around the clock. It's the difference between delegating to a team and pair-programming with them.
What Is VM Direct Access?
When you deploy a Synthcore project, we provision a dedicated AWS EC2 instance running Ubuntu 24.04. Your AI agents run on that VM 24/7 — writing code, running tests, filing bugs, and pushing commits to your GitHub repo.
VM direct access means you can SSH into that same VM and work in the same environment as your agents. You're not using a separate development machine — you're pair-programming with your AI team on shared infrastructure.
How It Works
- Connect via SSH — use the credentials provided in your Synthcore dashboard
- You're in the same repo — your VM clones your GitHub repo; agents and you share the same filesystem
- Run Claude directly — invoke Claude Code CLI manually for interactive sessions
- Use packaged skills — Synthcore's toolkit system gives you access to skills your agents already use
ssh -i ~/.ssh/synthcore-key ubuntu@your-vm-ip
cd /opt/synthcore/projects/your-project
claudeWhat Can You Do with Direct Access?
Pair-Program with Your Agents
Your agents are always working. When you log in, you can see exactly what they're doing — which files they're editing, which tests they're running, which PRs they're opening.
# See what your agents shipped recently
git log --oneline -10Run Interactive Claude Sessions
When you want to drive the keyboard yourself, invoke Claude Code directly. Your session runs alongside the autonomous agents — no conflict, no overwriting.
claude
# Now you're pair-programming with your AI teamAccess Shared Tools and Skills
Synthcore packages developer tools as skills — the same ones your agents use. On the VM, you have access to the full toolkit:
/security-scan— run OWASP dependency checks/scaffold— add boilerplate for new features- Git hooks via pre-commit toolkit
- Docker and container tools
Debug in Real Time
When an agent hits a tricky bug, you can log in, inspect the state of the repo, run the same tests the agent ran, and pick up where they left off. No need to wait for the next agent cycle.
Why Vibe Coders Love This
Vibe coding is about staying in flow — using AI to move fast without getting bogged down in boilerplate or context-switching. VM direct access keeps you in the loop without breaking your flow.
You're not babysitting agents. You're collaborating with them. When you want to drive, you SSH in. When you want to delegate, you close the terminal and let the agents run.
You see what they're building. Real-time visibility into agent activity means you're always aware of the current state of your project — no surprises when you open GitHub.
You can override anytime. If an agent is going down the wrong path, you can step in, correct course, and hand it back.
Security and Isolation
Each customer VM is isolated:
- No shared infrastructure — your VM is dedicated to your project
- Firewall-locked — only your IP can SSH in (security groups restrict port 22)
- API keys isolated — your model API keys stay on your VM, not in shared services
- Agent actions are auditable — all agent commits, test runs, and file changes are tracked in your audit log
Getting Started
VM credentials are available in your Synthcore dashboard under Project Settings → Access.
You'll need:
- An SSH key pair (or use the one Synthcore generates for you)
- Your VM's public IP address
- The password or key passphrase set during project creation
For full setup instructions, see the Getting Started guide.
What's Next
Once you're comfortable with direct access, explore the packaged skills — the same toolkit system your agents use to add tests, CI pipelines, and security scans in minutes.
Ready to pair-program with your AI team? Deploy your first project and get VM access in under 10 minutes.