Organize the knowledge of an AI system
Preserve meaning, reasons and direction for the people who must understand and evolve the system.
When architecture, documentation, research and explanations are scattered or confused with operational status.
Competence · Knowledge and continuity · Public edition 0.3.2
A team chose local processing because of a project condition. The knowledge base preserves reason, source and conditions.
The key relationship. Builds source-linked understanding and distinguishes what the system is, what it has shown and what it may become.
An illustrative situation to adapt to your work. Read the complete cases ↓
You supply the context: research, a service, a creative project or a team’s work. Examples make the method visible; adaptation preserves the relationships that make it work while understanding the purpose, sources, people and means of your own assignment.
Tell your AI what you want to achieve, what already exists and a practical situation to address. Ask how this competence changes the work and which knowledge from your domain should participate. MAIOS provenance identifies the shared knowledge; you can use it directly in your AI or compose it with an existing kernel.
Adaptation examples in other contexts ↗Project status says where the work stands. System knowledge also preserves why it is built that way, what its capabilities mean and which directions remain open.
This competence keeps that knowledge reachable from its sources. Guides, documentation, research and presentations can use it without reconstructing the whole system every time.
Retain knowledge when it changes identity, meaning, architecture, method, direction or use. Start from an attributable source, work-derived learning or observed behavior. Preserve what was understood and why it matters to future builders or readers. Do not accumulate every message, log or implementation detail merely because it is available.
Attribute the person's formulations, AI interpretations, observed capabilities and research hypotheses. A development intention remains a direction until its realization is evidenced. An observed result concerns the version and case actually exercised. A derived guide does not retroactively become the source of a concept.
Use an existing knowledge base or guide and connect detail to relevant sources. A short entry supports current work; deeper knowledge preserves reasons and relations. An index or graph can help when it makes a relation visible, but its shape does not exhaust the system or require loading everything at each use.
When an event changes understanding, preserve its source, the earlier interpretation, the new relation and its consequence. Update the knowledge that must continue from that difference. Superseded formulations retain provenance. If distinct objects emerge, separate them rather than forcing them under one name. Leave unresolved what the sources cannot establish.
Documentation, user guidance, presentations and research can draw on the same knowledge for different purposes. Supply the relevant relation and its status without inventing capability or maturity. Knowledge stewardship does not publish, establish scientific validity or automatically change a product or runtime.
Notice where reentry or explanation reconstructs knowledge that should already be reachable. Add missing knowledge, repair discovery when it exists, and correct compression when it lost a decisive reason. A new file makes knowledge available; later non-identical use shows whether it helps understanding and choice.
These cases illustrate situations, causal distinctions and choices you can adapt to your work.
Italiano. The following cases illustrate how the method is used.
A team chose local processing because of a project condition. The knowledge base preserves reason, source and conditions. A new collaborator understands when to reuse the choice and when a different context may call for another solution.
A presentation confuses a demonstrated capability with a desired one. Connect each to its source and revision. The text can explain both without presenting a direction as an already available function.
A system must be used and studied. Its guide receives instructions grounded in observed behavior; its research text receives hypotheses, conditions and evidence. The projections share meaning without becoming the same document.
Use SKILL.md or START_CHAT_EN.txt with the relevant sources and your actual context. Keep observations, interpretations and what remains open distinguishable. Preserve the useful reason and continuation in the project’s guide.
Builds source-linked understanding and distinguishes what the system is, what it has shown and what it may become.
Keep a source-linked knowledge base for system meaning, direction and reusable learning. Preserve the source occurrence, present understanding, why it matters, current or prospective consequences, possible consumers and consequential unknowns. Operator thesis, competence learning, verified incarnation, research relation and later projection remain distinguishable. A projection cannot retroactively become the source; a direction does not become a present fact through documentation.
Use taxonomy and topology as discovery relations when terms or known instances matter. A term index can distinguish kernels, competences, invariants, methods and projections without closing an unrepresented relation. A topology census gives qualified states; a graph may expose useful relationships without becoming the system itself. Inventory scattered research deposits by actual object and claim status before centralizing a corpus. Do not merge unrelated papers because they share an author or folder.
A Kernel Doc can explain system meaning, a guide operational use, business a qualified value and a paper a research argument with evidence and alternatives. These consume the knowledge in their own forms; none must be generated after every change. A small source-linked condensate can reduce reentry effort while the full corpus remains available. This overlap does not make a paper a runtime dependency or permission controller. Preserve availability of the condensate separately from later evidence of improved comprehension or latency.
Update the closest meaningful section rather than appending every log or transcript. Preserve earlier claims with their temporal lineage when later knowledge supersedes them. If the same difference changes both operational continuation and durable understanding, update their distinct owners without copying entire contents. The knowledge-stewardship method itself learns when preservation loses a meaningful relation.
Preserve the system's meaning, reasons and direction for people who need to understand and evolve it. Keep knowledge connected to sources and distinguish what the system is, what it has shown and what it intends to become.
Use this competence when architecture, documentation, research and explanations are scattered or confused with operational state. State says where a project stands; system knowledge also preserves why it has that form, what capabilities mean and which directions remain open. Guides, research, documentation and presentations can then use that knowledge without reconstructing everything.
Preserve knowledge when it changes identity, meaning, architecture, method, perspective or use. Begin from an attributable source, learning from work or observed behavior. Keep what became understandable and why it matters to someone building or explaining the system. Availability alone is no reason to accumulate every message, log or implementation detail.
Attribute the person's formulations, AI interpretations, observed capabilities and research hypotheses. A development intention remains a direction until its realization is evidenced. An observed result concerns the version and case actually exercised. A derived guide does not retroactively become the concept's source. Keep useful relations together without imposing compulsory classification stages.
Use an existing knowledge base or guide and connect detail to pertinent sources. A short entry orients current work; extended knowledge preserves reasons and relations for deeper reading. Make definitions, decisions and open questions accessible according to real need. An index or graph can expose useful relations without exhausting the system or requiring everything to be loaded each time.
When an event changes understanding, preserve source, earlier interpretation, new relation and consequence. Update the knowledge that must continue from that difference. An old formulation can be superseded without losing provenance. Separate distinct objects when they emerge instead of forcing them under one name. Keep unknowns explicit when sources do not resolve them.
Documentation, usage guides, presentations and research can use the same knowledge for different purposes. Give each its pertinent relation and state without inventing capability or maturity. Public explanations follow selected data and authorization. Knowledge maintenance does not itself publish, establish a scientific thesis or change the product or runtime. Those surfaces retain their own responsibility for effects.
Notice when reentry or explanation reconstructs knowledge that should already be reachable. Integrate absent knowledge, reconnect an entry that failed to reach it or correct compression that lost a decisive reason. A new file makes knowledge available; a later, different use shows whether it helps understanding and choice.
This competence preserves durable system understanding. Individual task continuity and operational state have different objects.
It can support a meta relation by making reasons for forming other capabilities available. An autological relation appears when it applies that understanding to its own knowledge selection and organization, such as correcting a summary that hid the difference between current capability and direction. This does not imply an autonomous monitor or formal proof of the system.
Knowledge evolves through sources and consequences. Keep useful lineage while leaving history that changes no present relation outside daily reentry.
Begin from available context and sources. Exercise the method in chat or connect actual tools and persistence. External actions retain their mandate. Instructions support use; behavior and value require observation in real work.
Read OPERATING_GUIDE.md as part of this method when forming the work. It carries the practical knowledge, decision relations and available means for this capability. The complete chat starter includes this knowledge and worked examples, so it remains usable without private sources. Preserve the public source identity and adapt through the recipient's actual means.
This competence preserves durable system understanding. Continuity of an individual task and control of operational state have different objects.
It can support a meta relation by preserving reasons used to form other capabilities. An autological relation appears when it applies that understanding to its own selection and organization of knowledge, such as revising a summary that blurred current capability and direction. This implies neither an autonomous monitor nor a formal proof of the system.
Competence, meta and autology: the conceptual guide ↗Knowledge evolves through sources and consequences. Preserve useful lineage while keeping history that does not change current work outside the everyday entry path.
Attach or paste START_CHAT_EN.txt, which contains the complete method, and add your context. The AI can work in the conversation's language. Retain the result and useful guidance if you want to resume in another chat.
Make the competence folder readable and have the agent read SKILL.md for the relevant work. Connect real sources and tools according to the environment's instructions. The kit does not install accounts, memory or recurring execution.
With an existing kernel, connect the method to the competence or guide already responsible for the work. Without one, apply the instructions directly and preserve useful knowledge in the project or a reusable note.
Read or copy the complete English starter and add your context. English is the kit’s primary language; Italian counterparts are included.
