DALIA

A containerized launcher for multi-agent systems written in DALI, powered by SWI-Prolog.

---

Pre-requisites

1. Install Docker
2. Clone DALI
   • Check the compatibility table:
     DALI–DALIA Compatibility
   • Example:

     git clone --branch swi-2026-03-03 --depth 1 https://github.com/AAAI-DISIM-UnivAQ/DALI_swi_prolog

---

Installation

1. Open a command shell (Command Prompt, Bash, etc.)
2. Clone DALIA
   • Check the compatibility table:
     DALI–DALIA Compatibility
   • Examples:

     git clone --branch swi-2026-03-03 --depth 1 https://github.com/AAAI-DISIM-UnivAQ/dalia_swi_prolog

     or

     git clone --depth 1 https://github.com/AAAI-DISIM-UnivAQ/dalia_swi_prolog

3. Navigate into the cloned repository:

   cd dalia_swi_prolog

Note: DALIA now uses SWI-Prolog (bundled in the Docker image). No Prolog license is required.

---

Usage

To launch the system, use the following command structure:

./run --dali <PATH-TO-DALI-DIRECTORY> --src <PATH-TO-MAS-DIRECTORY> [--token <OPENAI_API_KEY>]

• PATH-TO-DALI-DIRECTORY is the path to the cloned DALI repository.
• PATH-TO-MAS-DIRECTORY is the path to the multi-agent system project folder.
• Two example projects are provided: example and case_study_smart_agriculture.

---

Running with LLM Support (OpenAI)

To enable the LLM Bridge features (e.g. allowing agents to consult ChatGPT via acquire_knowledge/2), provide your OpenAI API key using the --token parameter:

./run --dali ../DALI_swi_prolog --src example --token sk-proj-xxxxxxxxxxxxxxxxxxxx

---

Running Offline (No LLM)

If you omit the --token parameter, the system will start in offline mode. Calls to the LLM service will receive a default "LLM disabled" response without contacting external APIs.

./run --dali ../DALI_swi_prolog --src example

---

Example Projects

Example 1: Emergency Response (example)

A multi-agent emergency response system with 9 agents:

Agent	Role
sensor	Detects events (fire, smoke, earthquake) and raises alarms
coordinator	Central coordinator, queries AI oracle, dispatches agents
logger	Logs all events and agent messages
communicator	Notifies people about emergencies
evacuator	Evacuates locations
manager	Dispatches appropriate equipment
responder	Responds at location with equipment
mary, john	People to be notified

Launch:

./run --dali ../DALI_swi_prolog --src example --token <YOUR_OPENAI_API_KEY>

Test — send a fire event:

In the user prompt, enter the following (one line at a time):

sensor.
me.
send_message(sense(fire, rome), me).

Expected flow:

1. sensor detects fire, logs the event, sends alarm to coordinator
2. coordinator queries the AI oracle, dispatches evacuator, manager, and communicator
3. evacuator evacuates the location, reports back
4. manager dispatches equipment (firetruck), reports back
5. communicator notifies mary and john
6. coordinator (when both evacuated + equipped) dispatches responder
7. responder responds at location with equipment, reports done

---

Example 2: Smart Agriculture (case_study_smart_agriculture)

A multi-agent precision agriculture system with 6 agents:

Agent	Role
soil_sensor	Receives soil readings (moisture, pH, field)
weather_monitor	Receives weather data (temperature, humidity, forecast)
crop_advisor	Queries AI oracle for advice, sends irrigation commands
irrigation_controller	Activates irrigation for specific fields
farmer_agent	Receives notifications about actions taken
logger	Logs all events

Launch:

./run --dali ../DALI_swi_prolog --src case_study_smart_agriculture --token <YOUR_OPENAI_API_KEY>

Test — send soil and weather data:

In the user prompt, enter the following commands (one line at a time, one command set at a time):

% Read soil data (moisture=20, pH=6.5, field=north_field)
soil_sensor.
me.
send_message(read_soil(20, 6.5, north_field), me).

% High temperature + drought forecast
weather_monitor.
me.
send_message(weather_update(42, 15, drought), me).

% Acidic soil in east field
soil_sensor.
me.
send_message(read_soil(50, 4.0, east_field), me).

% Frost warning
weather_monitor.
me.
send_message(weather_update(0, 80, frost), me).

Expected flow:

1. soil_sensor / weather_monitor receive data and forward to crop_advisor
2. crop_advisor queries the AI oracle for agricultural advice
3. crop_advisor sends irrigation commands and advisory notifications
4. irrigation_controller activates irrigation for the relevant field
5. farmer_agent receives notifications about actions taken