Components of an AI agent

You’ll use agent2.py, the Lesson 13 agent with two new parts: long-term memory in memory.json and an optional planning step. Error handling for tools is already built in.

The tools, the memory file and the step limit always behave exactly as shown. Which tools the model picks, and its wording, can differ a little on your laptop. On macOS or Linux, use python3, nano and cat instead of type.

1Run the agent and read its parts

Run this:

cd C:\ai-lab\agent
notepad agent2.py

Paste this whole block into agent2.py, then save:

import datetime
import json
import os
import urllib.request

MODEL = "llama3.2:3b"
QUESTION = "Nobody on the 2nd floor can print. Can you check?"
MEMORY = False        # True = load memory.json and add a remember tool
PLAN = False          # True = write a short plan before using any tools
APPROVE = True        # True = ask you before any tool that changes something
MAX_STEPS = 6         # stop after this many calls to the model
RULES = "You are an IT support agent. Check with the tools before you answer. Answer in two sentences."

# A pretend network, so the lab is safe to run on any laptop
SERVERS = {
    "web01": {"role": "intranet web server", "disk": 41, "services": {"nginx": "running"}},
    "file01": {"role": "file server", "disk": 97, "services": {"smb": "running"}},
    "print01": {"role": "print server for floors 1 to 3", "disk": 58, "services": {"spooler": "stopped"}},
}


def list_servers():
    return "\n".join(f"{name}: {s['role']}" for name, s in SERVERS.items())


def check_disk(server):
    return f"{server}: disk {SERVERS[server]['disk']}% used"


def check_service(server, service):
    state = SERVERS[server]["services"].get(service, "not installed")
    return f"{service} on {server}: {state}"


def restart_service(server, service):
    SERVERS[server]["services"][service] = "running"
    return f"{service} on {server} restarted, now running"


def load_memory():
    if not os.path.exists("memory.json"):
        return []
    with open("memory.json") as f:
        return json.load(f)


def remember(note):
    notes = load_memory()
    notes.append({"note": note, "saved": datetime.date.today().isoformat()})
    with open("memory.json", "w") as f:
        json.dump(notes, f, indent=2)
    return "Saved to memory."


FUNCS = {f.__name__: f for f in [list_servers, check_disk, check_service, restart_service, remember]}
CHANGES = {"restart_service"}      # tools that change something


def tool(name, description, **params):
    props = {p: {"type": "string", "description": d} for p, d in params.items()}
    return {"type": "function", "function": {"name": name, "description": description,
            "parameters": {"type": "object", "properties": props, "required": list(params)}}}


TOOLS = [
    tool("list_servers", "List all servers and what each one does."),
    tool("check_disk", "Show how full a server's disk is.", server="Server name, for example web01"),
    tool("check_service", "Show whether a service is running.", server="Server name, for example web01", service="Service name, for example nginx"),
    tool("restart_service", "Restart a service. Only use it when check_service shows a problem.", server="Server name, for example web01", service="Service name, for example nginx"),
]
if MEMORY:
    TOOLS.append(tool("remember", "Save a useful fact for future conversations.", note="The fact to save, in one sentence"))


def chat(messages, use_tools=True):
    body = {"model": MODEL, "messages": messages, "stream": False, "options": {"temperature": 0}}
    if use_tools:
        body["tools"] = TOOLS
    req = urllib.request.Request("http://localhost:11434/api/chat", data=json.dumps(body).encode(),
                                 headers={"Content-Type": "application/json"})
    with urllib.request.urlopen(req) as r:
        return json.load(r)


notes = load_memory() if MEMORY else []
system = RULES
if notes:
    system += "\n\nNotes from earlier conversations:\n" + "\n".join(f"- {n['note']} (saved {n['saved']})" for n in notes)
print(f"Model: {MODEL} | Tools: {len(TOOLS)} | Notes in memory: {len(notes)} | Plan: {'on' if PLAN else 'off'} | "
      f"Approval: {'on' if APPROVE else 'off'} | Step limit: {MAX_STEPS}")
messages = [{"role": "system", "content": system}, {"role": "user", "content": QUESTION}]
print("Question:", QUESTION)
if PLAN:
    ask = messages + [{"role": "user", "content": "Before you use any tools, write a numbered plan of at most 4 steps."}]
    plan = chat(ask, use_tools=False)["message"]["content"].strip()
    print("Plan:\n  " + plan.replace("\n", "\n  "))
    messages.append({"role": "assistant", "content": "My plan:\n" + plan})
for step in range(1, MAX_STEPS + 1):
    msg = chat(messages)["message"]
    messages.append(msg)
    calls = msg.get("tool_calls")
    if not calls:
        print("Answer:", msg["content"].strip())
        break
    for call in calls:
        name, args = call["function"]["name"], call["function"]["arguments"]
        print(f"Step {step}: {name}(" + ", ".join(f"{k}={v!r}" for k, v in args.items()) + ")")
        if name in CHANGES and APPROVE and input("  Allow this change? [y/n] ").strip().lower() != "y":
            result = "The user did not allow this change."
        else:
            try:
                result = FUNCS[name](**args)
            except Exception as e:
                result = f"Error: {e}"
        print("  -> " + result.replace("\n", "\n     "))
        messages.append({"role": "tool", "content": result, "tool_name": name})
else:
    print(f"Stopped: no answer after {MAX_STEPS} steps.")

Run this:

python agent2.py

When the script asks Allow this change?, type y and press Enter.

You should see something like this (the model’s words and choices may differ a little):

Model: llama3.2:3b | Tools: 4 | Notes in memory: 0 | Plan: off | Approval: on | Step limit: 6
Question: Nobody on the 2nd floor can print. Can you check?
Step 1: list_servers()
  -> web01: intranet web server
     file01: file server
     print01: print server for floors 1 to 3
Step 2: check_service(server='print01', service='spooler')
  -> spooler on print01: stopped
Step 3: restart_service(server='print01', service='spooler')
  Allow this change? [y/n] y
  -> spooler on print01 restarted, now running
Answer: The print spooler on print01 had stopped, so nobody could print. I restarted it and it is running again, so printing should work now.

The first line lists the components: model, tools, memory, planning and guardrails. Memory is off, so it takes three steps, like in Lesson 13.

✅ Checkpoint: The first line shows Tools: 4 and Notes in memory: 0, and the agent takes three steps.

2Turn on memory and teach it something

Change these two lines:

MEMORY = True
QUESTION = "Remember this: on print01 the spooler stops after Windows updates. Restarting it is safe."

Run this:

python agent2.py

You should see something like this (the model’s words and choices may differ a little):

Model: llama3.2:3b | Tools: 5 | Notes in memory: 0 | Plan: off | Approval: on | Step limit: 6
Question: Remember this: on print01 the spooler stops after Windows updates. Restarting it is safe.
Step 1: remember(note='On print01 the print spooler stops after Windows updates, and restarting it is safe.')
  -> Saved to memory.
Answer: Got it. I've saved that note, so next time I'll know the spooler on print01 can stop after Windows updates.

Memory on adds a fifth tool, remember. The model decided this message was worth saving and called it. Now look at the file it wrote:

Run this:

type memory.json

You should see something like this (your date will be today’s date):

[
  {
    "note": "On print01 the print spooler stops after Windows updates, and restarting it is safe.",
    "saved": "2026-10-19"
  }
]

The note is stored with the date it was saved. That date matters later in this lesson.

✅ Checkpoint: memory.json exists and holds one note about the print01 spooler.

3Ask again with memory

Change this line:

QUESTION = "Nobody on the 2nd floor can print. Can you check?"

Run this:

python agent2.py

You should see something like this (the model’s words and choices may differ a little):

Model: llama3.2:3b | Tools: 5 | Notes in memory: 1 | Plan: off | Approval: on | Step limit: 6
Question: Nobody on the 2nd floor can print. Can you check?
Step 1: restart_service(server='print01', service='spooler')
  Allow this change? [y/n] y
  -> spooler on print01 restarted, now running
Answer: The print spooler on print01 stops after Windows updates, so I restarted it. Printing should work again now.

One step instead of three. The note was in the system prompt, so the model went straight to the restart and even explained why.

Look closely, though: it didn’t check anything first. Keep that in mind for the second Break It lab.

✅ Checkpoint: Notes in memory: 1, and the agent restarts the spooler in one step.

4Add a planning step

Change these two lines:

PLAN = True
QUESTION = "Check every server and tell me what needs attention."

Run this:

python agent2.py

You should see something like this (the model’s words and choices may differ a little):

Model: llama3.2:3b | Tools: 5 | Notes in memory: 1 | Plan: on | Approval: on | Step limit: 6
Question: Check every server and tell me what needs attention.
Plan:
  1. List all servers.
  2. Check the disk on every server.
  3. Check the main service on every server.
  4. Report anything that needs attention.
Step 1: list_servers()
  -> web01: intranet web server
     file01: file server
     print01: print server for floors 1 to 3
Step 2: check_disk(server='web01')
  -> web01: disk 41% used
Step 2: check_disk(server='file01')
  -> file01: disk 97% used
Step 2: check_disk(server='print01')
  -> print01: disk 58% used
Step 3: check_service(server='web01', service='nginx')
  -> nginx on web01: running
Step 3: check_service(server='file01', service='smb')
  -> smb on file01: running
Step 3: check_service(server='print01', service='spooler')
  -> spooler on print01: stopped
Answer: Two things need attention: file01's disk is 97% full, and the print spooler on print01 has stopped. web01 looks fine, so ask me to restart the spooler if you want it fixed now.

The model wrote a four-line plan before using any tool, then followed it. In steps 2 and 3 it asked for three tools in one reply, so seven checks took only three model calls.

It also didn’t restart the spooler, because this question asked for a report, not a fix. The plan kept it on task.

✅ Checkpoint: A numbered plan appears first, and the answer names file01’s disk and the stopped spooler.