01
Create a scoped API token
Generate a token in client area with only the permissions your agent needs.
[ AI Automation ] 
Connect your AI agent to infrastructure through MCP
Connect once to Bullionet Hosted MCP and let your agent provision and operate servers automatically with strict permission scope and explicit safety checks.
AI Automation
Deploy 3 API servers in Paris, apply staging policy, and return a change summary.
[ Agent compatibility ]
Works with Claude Code, GitHub Copilot, Codex, and any agent with bash access
Any AI coding assistant with bash access can manage your infrastructure today via the REST API or bulli CLI. No special integration required.
[ Quick Setup ]
From zero to automated workflow in a few steps
No MCP infrastructure to run: Bullionet hosts it for you. Add endpoint + token and start controlled automation in minutes.
02
Connect Bullionet Hosted MCP
Copy your MCP endpoint from Client Area, add your API token, and your agent is ready.
03
Define guardrails
Set allowed actions, approval mode, and environment boundaries before first run.
04
Run production prompts
Start with concrete tasks like provisioning, scaling, and maintenance workflows.
[ MCP ]
Use your AI assistant with bulli CLI today
Grant your AI assistant a BULLI_TOKEN and it can provision, list, and operate servers using curl or bulli CLI. Bullionet Hosted MCP is in development.
Bullionet Hosted MCP, native one-click agent integration is in development.
API access, works today
Grant your AI assistant a BULLI_TOKEN and it can provision, list, and operate servers using curl or bulli CLI. Bullionet Hosted MCP is in development.
[ Concrete Use Cases ]
Production workflows teams actually run
Each workflow returns an executable plan, safety checks, and an auditable summary before sensitive actions.
Concrete Use Cases
Prepare a blue/green release for api-gateway in Paris: create 2 green nodes, run health checks, and propose cutover.
•Creates green nodes with release tags, location constraints, and bootstrap profile.
•Returns a health-check report (boot, SSH, app endpoint) before traffic switch.
•Requests approval for cutover and preserves rollback path to previous nodes.
[ Execution Safety ]
Automation with explicit boundaries
Keep automation fast but predictable with concrete controls at runtime.
Scoped permissions
Tokens and MCP context restrict what the agent can actually execute.
Approval gates
Use mandatory human approval on sensitive actions like reinstall or destructive changes.
Traceable execution
Capture command outputs and operation summaries for audit and incident reviews.
Launch AI automation without losing operational control
Connect your agent in minutes via Bullionet Hosted MCP, apply scoped permissions, and ship practical infrastructure workflows.
