FAQ / Understanding Knowledge Infrastructure

KAPRI stands for Knowledge Asset Package Reference Implementation.

KAPRI defines an open technical foundation for publishing knowledge states as Knowledge Asset Packages, exchanging them between systems and organizations, and validating them independently.

Its purpose is not to confirm the factual correctness of a knowledge state. KAPRI makes it possible to verify which organization a knowledge state originates from, which version was published, and whether the published representation has been modified since publication.

The reference implementation demonstrates how the specification can be implemented technically and how interoperability can be tested.

The broader goal is an open knowledge infrastructure in which responsible knowledge states remain clearly attributable and verifiable across system and organizational boundaries.

In short:

KAPRI provides an open technical foundation for the interoperable exchange of responsible knowledge states.

A knowledge network emerges when independent organizations publish, exchange, and use knowledge in an interoperable manner across organizational boundaries.

Each organization remains responsible for its own knowledge assets. 

Knowledge Asset Packages allow published versions of knowledge to be made available in a uniform, verifiable format.

Recipients can independently validate these packages and decide for themselves which organizations and publications they trust.

A knowledge network therefore does not require a central knowledge platform. It emerges through the regulated exchange between independent participants.

In short:

A knowledge network connects accountable knowledge publications from various organizations into an interoperable, federated exchange space.

A knowledge infrastructure provides the common formats, signature methods, and trust mechanisms that all participants in a knowledge network rely on. Without it, each organization would have to verify the origin and integrity of knowledge individually and inconsistently — a knowledge network would be practically unmanageable.

A knowledge asset is a well-defined, managed, and versioned collection of knowledge within an organization.

It may include, for example, rules, processes, methods, terms, models, or technical concepts. What matters is not the file format, but rather that the knowledge:

• has a defined subject matter and scope;
• is assigned to a responsible organization;
• is managed as a standalone asset;
• can be published in distinguishable versions.

A knowledge asset is therefore neither a single file nor a specific document. Files are merely possible representations of its content.

KAPRI publishes a specific version of a knowledge asset as a knowledge asset package. The package makes this version identifiable, exchangeable, and technically verifiable.

In short:

The knowledge asset is the knowledge for which an organization is responsible. The knowledge asset package is its published, interoperable representation.

A content item is a logically distinct unit within a knowledge asset.

Examples include:

• a term;
• a definition;
• a rule;
• a process step;
• a requirement;
• a model;
• a diagram;
• a data set.

A content item has its own identity and can be organized hierarchically within a composition or related to other content items.

It is independent of its technical representation: A content item can be represented by one or more files. Conversely, a single file can represent multiple content items.

In short:

A content item is an identifiable content building block of a knowledge asset—regardless of which file it is represented in.

Knowledge continues to evolve. New insights, experiences, or requirements may supplement, modify, or replace existing statements.

A state of knowledge describes the condition of a defined area of knowledge at a specific point in time. A version makes this state of knowledge uniquely identifiable.

An organization assumes responsibility for a state of knowledge by specifying:

• what content it includes;
• the scope for which it is intended;
• which version is being published;
• the date of publication;
• that this version has been approved by the organization.

A published state of knowledge remains unchanged. As knowledge evolves, a new state of knowledge emerges, which can be published as a new version.

Responsibility does not mean that the content is correct or binding for every purpose. Recipients decide for themselves whether to trust the organization and use the published body of knowledge.

In short:

A body of knowledge for which responsibility is assumed is a uniquely versioned representation of knowledge, assigned to an organization, at a specific point in time.

A document is one possible representation of knowledge. A knowledge asset, on the other hand, describes the knowledge itself—regardless of a specific file or form of representation.

A knowledge asset has:

• its own identity;
• a defined subject and scope;
• a responsible organization;
• traceable knowledge states and versions;
• a logical structure consisting of content items;
• a regulated publication and development process.

It can be represented by a document, multiple files, data, models, images, or other media. These representations may change without automatically creating a different knowledge asset.

In short:

A document represents knowledge. A knowledge asset turns this knowledge into an identifiable, accountable, and developable resource.

Knowledge becomes an organizational asset when it is not only present but is also deliberately defined, assigned responsibility for, and made available for long-term use.

To this end, an organization determines:

• which knowledge is valuable;
• what its subject and scope are;
• who is responsible for it;
• how it is maintained and further developed;
• which knowledge versions have been published;
• how its origin, integrity, and usability are preserved.

KAPRI supports this process by enabling a body of knowledge to be published as a uniquely identifiable, versioned, and verifiable Knowledge Asset Package.

This makes knowledge discoverable, traceable, shareable, and usable independently of individual people or systems. It can be specifically incorporated into processes, decisions, products, and cross-organizational collaboration.

Whether a Knowledge Asset can additionally be classified as a balance-sheet-eligible asset depends on the applicable legal and economic criteria.

In short:

Knowledge becomes an asset when it is made permanently usable as a managed resource and creates recognizable value for an organization or a knowledge network.

A Knowledge Asset Package, or KAP for short, is the published and verifiable representation of a specific version of a knowledge asset.

It links the actual content to the information required for its unique identification:

• Which knowledge asset is being represented?
• What version is it?
• Which organization is responsible?
• Which files are included?
• How is the content structured?
• Have they been modified since publication?

To this end, a KAP contains a manifest, a packing list, at least one composition, and the associated files.
The digital signature and cryptographic hashes enable recipients to technically verify the origin and integrity of the content.

In short:

A Knowledge Asset Package transforms a collection of files into an identifiable, structured, and verifiable knowledge publication.

ZIP, Git, OKF, and KAP address different problems.

A ZIP archive bundles files and makes them easier to transfer. However, it does not specify which published version of a knowledge asset the files represent or which organization is responsible for them.

Git manages changes and supports collaboration during the creation process. However, a repository is not a uniform, immutable publication format.

Google’s Open Knowledge Format organizes knowledge as Markdown files with YAML metadata. It is easy to read, Git-friendly, and particularly well-suited for knowledge catalogs and AI agents. However, OKF deliberately specifies only a few requirements for the content model.

A KAP, on the other hand, defines a verifiable publication. It standardizes:

• identity and version;
• responsible organization;
• logical structure;
• complete file inventory;
• cryptographic integrity;
• digital signature;
• optionally encrypted content and recipient-specific key distribution.

The formats can be combined:

• Git manages the creation process.
• OKF can structure the knowledge content.
• ZIP can simplify transport.
• KAP defines the binding publication and its technical verifiability.

For example, a KAP can contain a complete OKF bundle as its content.

In short:

    - Git manages changes.
    - OKF structures agent-readable knowledge.
    - ZIP packages files.
    - KAP publishes a verifiable version of the knowledge.

A composition describes the logical structure of a knowledge asset—regardless of how its content is stored in files.

It defines:

• which content items belong to the knowledge asset;
• how they are organized hierarchically;
• what relationships exist between them;
• which package files represent them.

Content items can form not only a hierarchy but also a network. Typed relationships can be used, for example, to express dependencies, references, prerequisites, or contextual connections.
A content item can be represented by multiple files, and a file can contribute to multiple content items.

In short:

The packing list specifies which files are present. The composition describes their logical order—as a hierarchy, a network, or a combination of both.

Integrity refers to the fact that a package is complete and has not been altered without detection since its release.

KAPRI verifies integrity at multiple levels:

• The packing list contains hash values and sizes of all package files.
• The manifest contains hash values of the packing list and the compositions.
• The digital signature protects the manifest and, thereby, the entire resulting chain of integrity.

This makes it possible to detect, for example, modified, corrupted, missing, or swapped files.

A technically intact package therefore still corresponds to the signed release version.

However, integrity does not mean that the content is factually correct, up-to-date, or trustworthy. This assessment remains the recipient’s own decision. 

A digital signature cryptographically links a knowledge asset package to its producer’s certificate. It makes it possible to verify which certificate was used to sign the package and whether it has been altered since then.

When transferring knowledge across organizational and system boundaries, it thus provides independent proof of origin and integrity.

Whether the recipient trusts the producer and uses the published body of knowledge remains their own decision.

In short:

The digital signature makes the origin and integrity of a transferred body of knowledge technically verifiable.

A digital identity enables an organization to identify itself during electronic exchanges.

To do so, it receives a certificate. This certificate contains the organization’s name and a public key. Using the corresponding private key, the organization can, for example, sign a Knowledge Asset Package.

The certificate chain shows who issued the certificate:

Root CA
   ↓
Intermediate CA
   ↓
Producer

The recipient can verify every link in this chain. This makes it possible to determine whether the producer certificate was actually issued by the specified certification authorities.

In KAPRI, the producer certificate identifies the signer of a package. The recipient certificate specifies for which recipient file keys are encrypted.

In short:

The certificate digitally identifies a participant. The certificate chain provides a traceable record of who has verified this identity.

Validation checks the objective technical properties of a Knowledge Asset Package.

For example, it answers the following questions:

• Is the package complete?
• Do the hash values match?
• Is the digital signature valid?
• Is the certificate chain technically correct?

Trust, on the other hand, is a decision made by the recipient. It answers questions such as:

• Do I trust the responsible organization?
• Is the published body of knowledge suitable for my purpose?
• Do I accept the certification authority used?
• Am I allowed or willing to use the content?

A package can be technically complete and unaltered even if the recipient does not trust its source. Conversely, trust in an organization does not make a corrupted or invalidly signed package technically valid.

In short:

Validation establishes technical facts. Trust determines the meaning the recipient assigns to them.

Interoperability means that different organizations and systems can exchange knowledge without having to develop a separate technical solution for each connection.

To achieve this, all parties involved must understand the same basic rules:how a body of:

• knowledge is identified and versioned;
• how content and files are structured;
• how origin and integrity are verified;
• how encrypted content is made available.

KAPRI defines these common rules. This enables a system to generate a Knowledge Asset Package that can be read and validated by an independent system.

In short:

Interoperability makes knowledge exchangeable and verifiable across organizational and system boundaries on a common basis.

Within a system, the meaning, structure, and origin of knowledge are often understood through local processes. When exchanging information with other organizations, this shared context is missing.

Without shared rules, important questions remain unanswered:

• What body of knowledge was transferred?
• Who is responsible for it?
• What content is included?
• How is it structured?
• Have any changes been made since it was published?
• How can the receiving system process the information?

Shared rules make this information unambiguous and machine-readable. This allows independent systems to exchange, verify, and classify knowledge without having to standardize their internal structures.

In short:

Shared rules ensure meaning, accountability, and traceability when knowledge crosses its original system boundary.