Glossary
Status: v0.1 (Draft)
Type: Normative (definitions); Informative notes where marked
Normative language: The key words MUST, MUST NOT, SHOULD, SHOULD NOT, and MAY are to be interpreted as described in RFC 2119.
AIC
Addressable Intelligence Commons (AIC) is a vendor-neutral, standards-oriented effort defining minimal shared semantics and declarations for interoperable interactions among addressable intelligent actors, independent of transport.
Actor
An actor is an identifiable participant that can receive messages and produce responses in accordance with AIC semantics.
An actor MAY be software, a service boundary, a human-mediated endpoint, or a composite system, provided it can publish an Actor Description and behave consistently with any conformance claim it makes.
Actor Description
An Actor Description is a published, retrievable document that describes an actor’s interoperability-relevant properties.
In AIC v0.1, an Actor Description MUST include: - actor metadata (including an identifier), - a Capability Declaration, and - a Governance Declaration.
Act
An act is the sender-declared communicative intent of a message, expressed using the AIC act vocabulary.
Acts are used to make interactions predictable across implementations without requiring a shared application schema.
Act label
An act label is the explicit representation of an act in a message (e.g., a field value, header value, or other substrate-specific marker).
An actor MAY infer acts as a convenience, but MUST support explicitly labeled acts for conformance.
Act vocabulary
The act vocabulary is the set of acts required for Layer 1.0 conformance in AIC v0.1:
INFORMASKREQUESTPROPOSE
Capability Declaration (Layer 1.1)
A Capability Declaration is the required portion of an Actor Description that states what the actor supports for interoperability, including at least:
- supported acts,
- supported representations (input/output),
- interaction modes, and
- material limits (or unspecified).
Capability declarations are declarations of intended support and MUST be treated as guidance rather than guarantees.
Conformance claim
A conformance claim is a public statement that an actor conforms to AIC v0.1 for a specified contiguous supported layer range starting at Layer 1.0 (e.g., 1.0–1.3).
Continuity artifact (Layer 1.5)
A continuity artifact is a durable, referenceable object used to maintain continuity across time and across asynchronous or multi-step interactions.
In AIC v0.1, continuity artifacts include commitments and receipts.
Commitment (Layer 1.5)
A commitment is a continuity artifact indicating that an actor has undertaken an obligation to perform some work under stated conditions.
Commitments are optional unless Layer 1.5 conformance is claimed.
Coordination semantics (Layer 1.4)
Coordination semantics are minimal conventions used to relate messages and responses across time, retries, and multi-step exchanges (e.g., correlation identifiers; optionally idempotency if an actor claims it).
Layer 1.4 is optional unless claimed.
Correlation identifier
A correlation identifier is a sender-provided identifier used to associate responses (and later related messages) with an initiating message or interaction thread.
If an actor claims Layer 1.4 conformance and supports correlation identifiers, it MUST echo the identifier in responses when provided.
Declaration
A declaration is published information about an actor’s semantics, capabilities, or governance posture intended to enable interoperability.
In AIC v0.1, the primary declarations are the Capability Declaration and Governance Declaration within the Actor Description.
Deferred interaction / DEFERRED
A deferred interaction is one in which the actor does not complete processing within the initial exchange and indicates that completion may occur later.
DEFERRED is a disposition indicating that the actor has not completed the requested work and that a later completion signal or retrieval may be required (as documented by the actor or a relevant profile).
Disposition
A disposition is the actor’s response classification for a message, chosen from the AIC disposition vocabulary.
Each response MUST be classifiable as exactly one disposition.
Disposition vocabulary
The disposition vocabulary is the set:
ACCEPTEDREFUSEDDEFERREDCOMPLETEDFAILED
Error category
An error category is a stable label indicating why an actor failed to complete processing (FAILED). In AIC v0.1, refusal and error categories are published together as part of the Governance Declaration.
Failed / FAILED
FAILED is a disposition indicating that processing did not successfully complete due to an error condition.
A FAILED response MUST include an error category drawn from the actor’s published refusal/error categories (or a clearly versioned successor set).
Governance Declaration (Layer 1.3)
A Governance Declaration is the required portion of an Actor Description that states the actor’s published governance posture relevant to interoperability, including at least: - retention posture, - refusal/error categories, - authority boundaries, and - human escalation posture (if applicable).
AIC does not define enforcement; governance declarations are transparency artifacts.
Human escalation
Human escalation is the actor’s declared posture that some interactions may be reviewed or handled by a human under specified circumstances.
If an actor declares human escalation, it SHOULD describe when it may occur at a coarse level.
Idempotency (Layer 1.4, optional)
Idempotency is a property of an interaction in which repeating the same request (as defined by an idempotency key or equivalent mechanism) does not cause unintended repeated effects.
AIC v0.1 does not mandate idempotency. If an actor claims idempotency support, it MUST document how it is expressed and what guarantees are provided.
Interaction mode
An interaction mode is the actor’s declared timing pattern for exchanges. In v0.1:
- sync: completion is typically provided in the initial response
- deferred: completion may occur later (often via later retrieval or callback, as documented by the actor/profile)
Layer
A layer is a scoped, versioned set of AIC requirements and vocabulary.
AIC v0.1 defines Layers 1.0 through 1.5 as an incremental stack. Conformance claims MUST specify a contiguous supported range starting at Layer 1.0.
Message
A message is a unit of communication sent to an actor that includes: - an act label, and - content in some representation.
AIC does not mandate a transport or wire format for messages.
Profile
A profile is a substrate-specific mapping of AIC semantics onto an existing protocol or environment (e.g., HTTP+JSON).
Profiles are informative unless explicitly declared normative by an actor for interactions with that actor. AIC v0.1 does not require any profile.
Qualifier (Layer 1.2)
A qualifier is an optional, standardized modifier that refines how an act should be interpreted (e.g., priority, time constraints, output preferences).
Layer 1.2 is optional unless claimed. If claimed, supported qualifiers MUST be published and implemented as specified.
Receipt (Layer 1.5)
A receipt is a continuity artifact indicating the final outcome of a previously accepted or deferred interaction, typically referencing the initiating request and/or commitment and stating a terminal disposition (COMPLETED or FAILED) with associated result/error details.
Receipts are optional unless Layer 1.5 conformance is claimed.
Refusal category
A refusal category is a stable label indicating why an actor refused a message (REFUSED) or failed to complete it (FAILED).
Actors conforming to Layer 1.3 MUST publish their refusal/error categories and MUST only emit categories from the published set (or a clearly versioned successor set).
Refused / REFUSED
REFUSED is a disposition indicating that the actor will not perform the requested work (or cannot accept the interaction as stated), typically due to policy, authorization, capability, or constraints.
A REFUSED response MUST include a refusal category drawn from the actor’s published refusal/error categories (or a clearly versioned successor set).
Representation
A representation is the form/encoding of message content (e.g., plain text, JSON).
Representations SHOULD be identified using established media type strings where applicable.
Retention posture
A retention posture is the actor’s coarse, published statement about whether and how interaction data (inputs/outputs and related artifacts) may be retained, and for what general purpose and duration.
AIC does not mandate specific retention behavior; it requires transparency via declaration for actors claiming Layer 1.3 conformance.
Result
A result is the substantive output of processing a message when the disposition is COMPLETED.
AIC does not define result schemas; results are representation-dependent and may be domain-specific.
Supported layer range
A supported layer range is the contiguous prefix of AIC layers an actor claims to implement, starting at Layer 1.0 (e.g., 1.0–1.3, 1.0–1.5).
Terminal disposition
A terminal disposition is a disposition that indicates the interaction has reached an end state. In v0.1, terminal dispositions are:
- COMPLETED
- FAILED
- REFUSED
ACCEPTED and DEFERRED are non-terminal.
updated_at
updated_at is the Actor Description metadata field indicating when the description was last updated (or its version marker). It MUST change when materially relevant declared semantics change.
Vocabulary
A vocabulary is a controlled set of terms with defined meanings used to achieve interoperability.
In AIC v0.1, the primary vocabularies are the act vocabulary and disposition vocabulary, plus any published refusal/error categories and supported qualifiers.