Agents

Build specialized automation units with Python

What are Agents?

Agents are the fundamental building blocks of ViksaAI. Each agent is a specialized Python module designed to perform a specific task—whether it's querying a database, calling an API, processing data, or interacting with external systems.

Agents are designed to be invoked by the execution engine: when you give a goal in natural language, the Agent Executor (Think→Act→Observe) decides which agents to call and in what order at runtime. No pre-defined workflows—the path is discovered as the executor runs.

Key Features

Python Native

Write agents in pure Python with any libraries you need.

Typed I/O

Define strongly-typed inputs and outputs for reliability.

Multiple Endpoints

Each agent can expose multiple callable functions.

Secret Management

Securely inject API keys and credentials at runtime.

Build & Deploy

Automated build pipeline with dependency management.

Marketplace Ready

Publish agents to share with the community.

Agent Lifecycle

1

Create

Define inputs, outputs, and write code

2

Build

Install dependencies and validate

3

Deploy

Make available for execution

4

Execute

Invoked by the executor when you run a goal

Build Status

Agents go through a build process to validate code and install dependencies:

pendingin_progresssuccessfailure

Deploy Status

After a successful build, agents can be deployed:

not_deployedscheduleddeployingdeployeddeployment_failed

Guides

Creating Agents

Generate with AI or manual: configure, async Python, build, and deploy.

15 min read

Auth & credentials

Auth methods, vault-backed secrets, and ViksaAuth in main.py.

6 min read

Agent Types

Cloud (platform-managed) vs Secure (your Chrona workers).

8 min read

Endpoints

Map endpoints, inputs, outputs, and execution hints per endpoint.

10 min read