Guide · Agent skill
How to use the BDI Mental State Modeling agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The bdi-mental-states skill teaches an AI agent to represent its internal state as Belief-Desire-Intention (BDI) structures written in RDF/Turtle. It is aimed at developers building BDI or multi-agent systems, researchers on explainable or neuro-symbolic AI, and engineers who want semantic web data feeding agent decisions. After reading this guide you will know what the skill changes, how to install it, how to trigger it, and when it is the wrong tool.
BDI Mental State Modelingmuratcankoylan/Agent-Skills-for-Context-Engineering · MIT Download the skill Quick start
- 1Download the ZIP. Download the bdi-mental-states skill ZIP from this page.
- 2Unzip the skill. Unzip the archive so the bdi-mental-states folder contains SKILL.md and the references folder.
- 3Place the folder. Move the folder into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ directory.
- 4Or upload in the Claude apps. Alternatively, upload the skill under Settings → Capabilities → Skills.
- 5Ask a BDI question. Describe your task using terms like beliefs, desires, intentions or RDF-to-belief conversion.
- 6Check the output. Review the generated Turtle or SPARQL, including prefixes and class names, before using it.
What the skill makes the agent do
Once the skill is active, the agent models its mental state formally. Beliefs are tied to world state references, desires are linked back to the beliefs that motivate them, and intentions are linked to the desires they fulfil and the plans that specify them. Each mental entity can carry a justification, and mental states can have validity time intervals.
The skill also describes a Triples-to-Beliefs-to-Triples (T2B2T) pipeline: incoming RDF triples become belief instances, and after deliberation the results are projected back into RDF. The output is machine-readable linked data for multi-agent or semantic systems, not an ordinary chat answer.
- Beliefs: what the agent holds true, grounded in a world state
- Desires: what it wants to bring about, motivated by beliefs
- Intentions: what it commits to, fulfilling a desire and specifying a plan
- Justifications: the evidence or rule behind each mental state
- Validity intervals: time bounds for beliefs using TimeInterval instances
Install the bdi-mental-states skill
Download the skill ZIP from this page and unzip it. The skill folder contains SKILL.md and a references folder with four files: bdi-ontology-core.md, framework-integration.md, rdf-examples.md and sparql-competency.md.
For Claude Code, place the unzipped skill folder in your personal skills directory ~/.claude/skills/, or in a project's .claude/skills/ directory to scope it to one project.
In the Claude apps, you can instead upload the skill under Settings → Capabilities → Skills.
- Personal install: ~/.claude/skills/bdi-mental-states/
- Project install: .claude/skills/bdi-mental-states/ inside your project
Trigger the skill with natural requests
The skill activates from its description, so you do not need a special command. Ask in terms of the topics it covers: beliefs, desires, intentions, RDF-to-belief conversion, BDI ontologies, rational agency traces, or neuro-symbolic integration. Naming a framework such as SEMAS, JADE or JADEX also helps.
The skill tells the agent not to activate for adjacent work. General context-window questions belong to context-fundamentals, persistent user memory to memory-systems, topology decisions to multi-agent-patterns, and quality rubrics to evaluation.
- Convert this RDF sensor feed into agent beliefs with world state references.
- Model the intention to book a room as a BDI structure with its supporting justification.
- Make the reasoning behind this agent's decision traceable in Turtle.
- Show how a desire is generated from existing beliefs in a SEMAS rule.
Example: a belief, desire and intention chain
The skill's core pattern links the three mental states in both directions, so the agent can reason forward (what should it do?) and trace backward (why did it act?). The example below is taken from the skill's cognitive chain pattern.
:Belief_store_open a bdi:Belief ;
rdfs:comment "Store is open" ;
bdi:motivates :Desire_buy_groceries .
:Desire_buy_groceries a bdi:Desire ;
rdfs:comment "I desire to buy groceries" ;
bdi:isMotivatedBy :Belief_store_open .
:Intention_go_shopping a bdi:Intention ;
rdfs:comment "I will buy groceries" ;
bdi:fulfils :Desire_buy_groceries ;
bdi:isSupportedBy :Belief_store_open ;
bdi:specifies :Plan_shopping .What you need before you start
The skill lists no required packages, runs no scripts and makes no network calls, so it is only instructions and reference files for the agent. The output is Turtle and SPARQL text. To use that output you still need your own RDF tooling and, if you want execution, your own BDI framework such as JADE, JADEX or SEMAS. The skill does not provide these.
- Requires: nothing listed by the skill
- Runs scripts: no
- Uses network: no
- You supply: an RDF toolchain and, optionally, a BDI framework
Safety notes and tips
Because the skill runs no scripts and uses no network, its risk is mainly in what the agent writes. Review generated ontology patterns before loading them into a production system, especially the namespace and class names, since the examples use prefixes such as bdi: and ex:.
Ground beliefs in world state references instead of free text, and attach a justification to every mental entity. Use temporal intervals so conflicting beliefs can be checked with SPARQL filters.
- Check generated prefixes and class names against your own ontology.
- Keep the reference files open for framework-specific mapping details.
- Do not use the skill for everyday chat help; it adds formal structure with no benefit there.
What people use it for
- Converting an RDF sensor or notification feed into agent beliefs with provenance back to the source.
- Modelling an agent's decision to act as a traceable chain from beliefs to desires to intentions.
- Writing SPARQL temporal filters to find which beliefs were active at a given time.
- Mapping BDI constructs to SEMAS production rules for a multi-agent prototype.
Frequently asked questions
How do I install the bdi-mental-states skill?
Download the skill ZIP from this page and unzip it. Put the folder in ~/.claude/skills/ for personal use or in your project's .claude/skills/ directory. In the Claude apps, you can upload it under Settings → Capabilities → Skills.
Does the bdi-mental-states skill require anything else to run?
The skill lists no requirements, runs no scripts and uses no network. It provides instructions and reference files only. You will need your own RDF tooling and any BDI framework you plan to run the output in.
Is the bdi-mental-states skill free?
The skill is published under the MIT licence. Check the licence file in the source repository for the full terms.
What is the bdi-mental-states skill not good for?
It is not for general reasoning help, persistent user memory or conversation history, or evaluating agent quality and designing multi-agent topologies. The skill redirects those tasks to memory-systems, evaluation and multi-agent-patterns.
Which frameworks does the skill support?
The skill's guidance covers SEMAS, JADE and JADEX, with SEMAS rule translation patterns in its framework integration reference. It is aimed at developers building BDI or multi-agent systems.
Why does the output look like Turtle instead of a normal answer?
The skill is designed to produce machine-readable linked data for semantic or multi-agent systems. If you only want a plain-language explanation, a general assistant without this skill will suit you better.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use this skill if you are building BDI or multi-agent systems, doing neuro-symbolic or explainable AI research, or wiring RDF and SPARQL data into agent decisions. Skip it for everyday reasoning, memory of past conversations, evaluation rubrics or multi-agent topology design, since other skills cover those better.
Download the skill