Auto-generates OpenAPI manifest with policies + MCP tools
03
Deploy
One click to a production endpoint. Versioned, auditable.
No projects yet. Create your first one above.
AgentOS Sandbox
Policy enforcement sandbox. Test agent actions safely with risk scoring.
Quick Templates
Click a template to load example code
Code Editor
Policy Evaluation & Risk Analysis
Run code to see policy evaluation, risk score, and execution output.
Risk Score0
0-30: Safe30-70: Caution70-100: High Risk
Execution Output
Policy Reference
Action
Risk
Policy
shell/exec
80
Arbitrary code execution blocked
fs.read/write
75
File system access restricted
net/http
50
External calls allowed with limits
crypto
10
Cryptographic operations allowed
console.log
5
Logging always allowed
Stubbed APIs
Slack
Post messages, list channels
Active
GitHub
Create issues, list repos
Active
Stripe
Create charges, refunds
Active
Risk scoring
Action
Risk
Reason
shell
80
Arbitrary code execution
email
50
External communication
http
40
External network call
Receipts
Every execution returns a signed receipt. ed25519. Verifiable.
Receipt ID
Project
Action
Cost
Time
No receipts yet.
MCP Tools
Model Context Protocol tools generated from your workflows.
What are MCP Tools?
MCP (Model Context Protocol) tools let AI agents discover and call your workflows directly. Each tool includes a name, description, and JSON schema that agents use to understand inputs and outputs.