Skip to main content
The Agno provider integrates the Agno AI agent framework with Pragmatiks, enabling you to define, configure, and deploy AI agents and multi-agent teams as declarative resources.

​
What is Agno?

Agno is a Python framework for building AI agents with support for models, tools, knowledge bases, and memory. The Agno provider lets you manage the full agent lifecycle declaratively — from model configuration to Kubernetes deployment.

​
Setup

The Agno provider is a first-party provider maintained by Pragmatiks. Install it with: 
pragma providers push --deploy

​
Resources

ResourceTypeDescription
Agentagno/agentAI agent definition with model, tools, and knowledge
Teamagno/teamMulti-agent team with coordinated members
Modelsagno/models/anthropic, agno/models/openaiLLM configuration (Claude, GPT)
Toolsagno/tools/mcp, agno/tools/websearchAgent tool integrations
Knowledgeagno/knowledge, agno/vectordb/qdrant, agno/knowledge/embedder/openai, agno/knowledge/contentSemantic search and RAG
Memoryagno/memory/managerPersistent agent memory
Storageagno/db/postgresPostgreSQL storage for sessions and memory
Deploymentagno/runnerDeploy agents/teams to Kubernetes

​
Architecture

Resources form a dependency graph. A typical deployment looks like: 
Secret (API key)
  └─▶ Model (Claude/GPT)
        └─▶ Agent
              ├─▶ Tools (MCP, WebSearch)
              ├─▶ Knowledge (VectorDB + Embedder + Content)
              ├─▶ Memory (Manager + DB)
              └─▶ Runner (Kubernetes Deployment)
Changes propagate automatically. Updating a model triggers the agent and runner to rebuild.

​
Dependencies

The Agno provider depends on:
  • GCP provider — for gcp/gke clusters (required by agno/runner)
  • GCP provider — for gcp/secret (commonly used for API keys)