Skip to main content

Guides

Practical guides for building and deploying agents with AgentVisor™.

Overview​

These guides extend what you learned in the Tutorial, showing you how to apply those concepts in real-world scenarios. Each guide focuses on a specific task or pattern, providing step-by-step instructions you can follow immediately.

For deeper understanding of how AgentVisor works under the hood, see the Concepts section after completing the relevant guides.


Getting to Production​

Essential guides for deploying AgentVisor in production environments.

GuideDescription
Scaling and ResilienceHigh availability deployment and horizontal scaling patterns
Production DeploymentPre-deployment checklist and hardening
LicensingObtaining and activating a Manetu license key, including container and air-gapped deployments
Temporal ConfigurationComplete Temporal setup including Cloud, self-hosted, encryption, and namespaces
Checkpoint Sizing and LimitsPayload size ceilings, the config knobs that raise them, and how to tell when a thread is near one — edge-case reading, most workloads never need it
External Checkpoint StorageRaise checkpoint size ceilings with a pluggable storage backend (PostgreSQL, S3, YCQL, filesystem) — edge-case reading, most workloads never need it

Authentication & Security​

Configure authentication and authorization for your agents.

GuideDescription
OIDC SetupConfigure OpenID Connect authentication
Policy ConfigurationWriting effective authorization policies
Multi-tenancyTenant isolation patterns
Security MonitoringRuntime tracing for security monitoring and SIEM integration

Agent Development​

Patterns and techniques for building sophisticated agents.

GuideDescription
Human-in-the-LoopInterrupt and resume patterns
StreamingReal-time LLM output
Local Development SetupPython venv setup, --sandbox=none mode, interactive debugging
Agent LoggingStructured logging for agent code
Agent TracingDistributed tracing with Jaeger, Langfuse, Datadog, and other OTLP backends

Framework-Specific Guides​

Guides for specific agent frameworks. See Supported Frameworks for current support status.

GuideFrameworkDescription
Framework ProvidersAllProvider selection algorithm and schema discovery shared across every framework
LangGraph AgentsLangGraphBuilding multi-node graphs, checkpointing, state management
CrewAI AgentsCrewAIMulti-agent crews, crewai.yaml discovery, MCP tool integration
Google ADK AgentsGoogle ADKADK agents, session state routing, agent.yaml discovery
Strands AgentsStrandsExplicit provider selection, snapshot checkpointing, MCP tools, LiteLLM/Bedrock
Interactive ExamplesInteractive (exec mode)Running Claude Code, Aider, shells inside policy-enforced sandboxes

API Transports & Gateways​

AgentVisor supports multiple protocols for accessing agents, split by direction: Inbound Transports (external clients call your agents) and Outbound Gateways (your agents call external services). Each is independently enabled.

Inbound Transports​

GuideDescription
OpenAPI/RESTOpenAPI/REST transport (LangGraph Agent Protocol v0.2.0), enabled by default
MCP TransportExpose agents as MCP tools for Claude Desktop, Cursor
A2A TransportGoogle Agent-to-Agent Protocol for agent interoperability
A2A Agent CardsEnrich Agent Cards with tags, examples, and metadata

Outbound Gateways​

GuideDescription
MCP GatewayAccess external MCP servers from your agent code
A2A GatewayAccess external A2A agents from your agent code

Environment & Configuration​

Understanding project structure and configuration options.

GuideDescription
Environment ConfigurationUsing .env files for secrets and config
Project StructureWhat each file does, how it all fits together

Troubleshooting​

Running into problems? See the Troubleshooting guide for common issues and solutions.