Anvia Studio

Anvia Studio is a local browser console for building, running, and debugging Anvia agents and pipelines. Point it at the same runtime objects your application uses, then inspect their behavior without building a temporary interface or test harness.

Studio is designed for the development loop:

Run a prompt → inspect the response → review tools and traces → adjust the agent → run again

What Studio gives you

Studio discovers capabilities from the agents and pipelines you register. The browser can then expose the surfaces that apply to them.

Surface	What it helps you do
Playground	Stream a real agent run and keep related prompts in a session.
Agents	Verify identity, model, tools, context, memory, hooks, and runtime limits.
Tools	Inspect schemas and invoke a tool directly with validated arguments.
Traces	Follow model generations and tool calls for a Studio session.
Pipelines	Read a pipeline graph, run it, inspect logs, and replay saved input.
Knowledge	Inspect registered context, dynamic tools, and retrieval evidence.
Memory	Explore users, conversations, messages, and recorded run steps.
MCPs and sandboxes	Inspect connected MCP tools and active sandbox resources.
Status	Check registered targets, storage adapters, and enabled capabilities.

These are not mocked representations. Playground runs call the registered agent, tool execution uses the registered tool, and pipeline runs use the registered pipeline.

Where Studio fits

Studio and Lens answer different questions.

Studio	Lens
Build and debug agents locally.	Observe applications across environments.
Run prompts, tools, and pipelines interactively.	Investigate production traces, users, sessions, cost, and failures.
Keep rapid development feedback close to the code.	Retain telemetry and evaluate behavior over time.
Starts with runtime objects in the current process.	Starts with telemetry exported by an application.

Use Studio while changing the agent. Use Lens when the application needs durable observability, evaluation workflows, and operational investigation.

Start here

Page	What you will do
Install and setup	Add Studio to an Anvia project and prepare a local entry point.
Run your first agent	Register an agent, open the Playground, and inspect a real run.
How Studio works	Understand targets, discovery, execution, storage, and lifecycle.

Development boundary

Studio can execute agents, tools, and pipelines with the credentials and permissions of its process. Treat it as a trusted developer surface: bind it to a loopback address, keep provider keys on the server, and do not expose it directly to an untrusted network.

Start with Install and setup.