Agents

Ockam Agents are intelligent autonomous actors powered by large language models. They understand natural language, engage in conversations, and perform complex tasks. Each agent:

• Maintains a unique identity and long-running conversations
• Accepts customization with specific knowledge and capabilities
• Runs securely in an isolated environment
• Communicates with other agents and systems using Ockam’s secure messaging protocols

There are three key files that define an agent. First is the ockam.yaml file:

name: example001
pods:
  - name: main-pod
    public: true
    containers:
      - name: main
        image: main

The ockam.yaml file above defines:

• name: example001 - A unique name for this deployment zone
• pods - A list of pods to run in the zone
  • name: main-pod - The name of the pod
  • public: true - Makes the pod accessible from the internet
  • containers - The containers to run in the pod
    • name: main - Container name
    • image: main - References the image built from the Dockerfile located at images/<value>/Dockerfile

This minimal configuration creates a deployment that runs our agent code.

Next is the Dockerfile file:

FROM ghcr.io/build-trust/ockam-python:latest
COPY . .
ENTRYPOINT ["python", "main.py"]

The Dockerfile above:

• Uses the official Ockam Python base image which includes the Ockam SDK
• Copies the agent code (main.py and other files) into the container
• Sets the entrypoint to run main.py when the container starts

Finally is the main.py file:

from ockam import Agent, Node


async def main(node):
    await Agent.start(node, "You are Henry, an expert legal assistant", "henry")


Node.start(main)

The main.py file above:

• Imports the Ockam SDK
• Defines an async main function that starts an agent with a specific prompt
• Starts the node and agent

When you run ockam it will:

• Build the image from the Dockerfile
• Deploy the container to the Ockam serverless runtime
• Expose a public HTTP endpoint to the agent
• Create two local (i.e., on your workstation) portals to the agent’s HTTP endpoint and log output.

» ockam
    Deploying zone example001 in cluster 25df35de87aa441b88f22a6c2a830a17...

  ✔ Created a repository for the image main
  ✔ Built image main
  ✔ Pushed image main

  ✔ Deployed zone example001 in cluster 25df35de87aa441b88f22a6c2a830a17

    The http server on the main-pod is available at:
    https://25df35de87aa441b88f22a6c2a830a17-example001.ai.ockam.network

  ✔ Opened a portal to the outlet http on main-pod from tcp://localhost:8000
  ✔ Opened a portal to the outlet logs on logs-pod from tcp://localhost:3000

    ─────────────────────────────────────────────────
    Press Ctrl+C to stop all portals and exit

​

Logs

You can see the logs of your running python app on http://localhost:3000/logs/explorer .

​

HTTP Server

http GET \
"https://25df35de87aa441b88f22a6c2a830a17-example001.ai.ockam.network/agents"

http POST \
  "https://25df35de87aa441b88f22a6c2a830a17-example001.ai.ockam.network/agents/henry" \
  message="who are you"

http POST \
  "https://25df35de87aa441b88f22a6c2a830a17-hello.ai.ockam.network/agents/henry?stream=true" \
  message="who are you"