SuperAPI AI Directory

bdi-mental-states

Agents & models

View repository

Updated 10-16 05:48

About this skill

🧠 BDI Mental States (Cognitive Modeling)

This skill allows modeling the internal state of autonomous agents using the BDI (Belief-Desire-Intention) framework. It uses formal representations (RDF/Turtle) to ensure that the agent acts with logical coherence and purpose.

Mental Model: The agent is not reactive, it is deliberative. Its actions (Intentions) must be justified by its goals (Desires) and its perception of the world (Beliefs).

🚩 Fragility and Autonomy

  • Fragility: High. Inconsistencies between beliefs and desires can lead to logical paralysis or infinite loops.
  • Freedom: Medium in the definition of mental states, but Low in the structure of the base ontology.

🚀 When to Activate

  • When designing agents that require complex long-term reasoning.
  • To formalize decision-making in multi-agent systems.
  • When debugging agent behaviors that seem to "lose the thread" of their mission.
  • To implement a robust "Context Awareness" layer.

🧠 Core Concepts (BDI)

1. Beliefs

They represent what the agent "knows" or "perceives" about the world.

  • Format: RDF triples that describe facts (e.g. agent_a believes sky_is_blue).
  • Dynamics: They are updated through perception.

2. Desires

They represent the final states that the agent "would want" to reach.

  • Format: Potential goals (e.g. agent_a desires target_location_reached).
  • Dynamics: Motivated by beliefs.

3. Intentions

They represent the desires to which the agent has "committed".

  • Format: Executable plans or action commitments.
  • Dynamics: An intention persists until it is fulfilled or becomes impossible (based on beliefs).

🛠️ Implementation in Gemini Elite Core

Pattern: Turtle Representation

Gemini 3 is excellent at generating Turtle RDF. Use it to maintain a rigid mental state.

@prefix bdi: <https://w3id.org/fossr/ontology/bdi/> .
@prefix ex: <http://example.org/> .

ex:AgentG1 a bdi:Agent ;
    bdi:hasBelief ex:ServerIsDown ;
    bdi:hasDesire ex:ServerIsUp ;
    bdi:hasIntention ex:RebootServer .

ex:RebootServer bdi:fulfils ex:ServerIsUp ;
    bdi:isSupportedBy ex:ServerIsDown .

Deliberation Flow

  1. Perception: Updates the Beliefs graph.
  2. Desire Generation: Identifies discrepancies between Beliefs and the Ideal.
  3. Commitment: Selects the highest-priority Desire and turns it into an Intention.
  4. Planning: Generates the technical steps to fulfill the Intention.

🔗 Integrations

  • context-fundamentals: Basis for descriptions of mental states.
  • multi-agent-patterns: Coordination of intentions among several agents.
  • project-development: Application of BDI to software lifecycles.

📚 Internal References

  • references/bdi-ontology-core.md: Formal definition of classes and properties.
  • references/rdf-examples.md: Example library in Turtle for various scenarios.
  • references/sparql-competency.md: Queries to verify mental integrity.
  • references/framework-integration.md: Connection with JADE/Python/LAG.