EnRuSimple EnSimple Ru
Acts of Emergence

Glossary of Terms

This document provides definitions for the key concepts used in the Idea Protocol and its related systems.

  • Idea: A self-contained, stateful data triplet (schema, solution, context) representing a unit of knowledge. It's a persistent, computational primitive, not an ephemeral prompt.

  • Ideator: An Idea that accepts input, signified by a context message of type: "input". It acts as a function, transforming input to output.

  • Idea Transformer: A specialized Ideator that takes another Idea as its input.

  • Input Message: A context message containing a schema and input data. It defines the expected inputs for a Request, making it a reusable, function-like component.

  • Tool: A schema that defines a capability an agent can use. It is presented to an LLM as part of a request, acting as a structured interface for a potential action. The LLM activates the tool by generating a Call with specific parameters, which is then executed either latently by the LLM or explicitly by a registered code function (Activity).

  • Activity: An explicit, deterministic code function that implements the logic for a Tool. It is the mechanism for executing actions that require external API calls, database operations, or any task that cannot be handled by an LLM's latent space.

  • Branch: A named tag that partitions the visibility space, creating a parallel, isolated environment for development and experimentation. Associating an Idea with a branch is an act of publication.

  • Call: A concrete, executable instance of a Tool, with specific values for its params. It's an invocation-focused request for what _should be done_.

  • Cutoff Time: A timestamp that accompanies a resolution request, instructing the resolver to find the version of an Idea that was considered latest as of that specific moment in time.

  • Method: The mechanism of execution for a Call.

    • Explicit Execution: The Call's output is generated by deterministic code (an Activity).y an LLM.

    • Latent Execution: The Call's output is generated by an LLM.

  • Delegate: A protocol for isolating execution context. Invoked by a Call's _delegate property, it executes an Activity or a new Request in a "clean room" environment, with the _scopes property providing controlled access to the parent context.

  • Scope: A key that identifies a piece of context from the parent environment to be made available (scoped) to an execution. It can be used to focus an LLM's attention in a Latent Inline execution or to construct the entire context for a Delegate Scope execution. Controlled by the _scopes property.

  • Loop: A sequence of Requests aimed at achieving a goal. The agent continues to invoke Requests, process the resulting Calls, and feed the output back into the context of the next Request until no more Calls are generated.

  • HITL (Human-in-the-Loop): The practice of inserting a human checkpoint into an automated process. In the context of the Execution Loop, this refers to the opportunity for a user to approve, deny, or modify an agent's proposed Calls before they are executed.

  • Hierarchical Versioning: A versioning scheme where versions are dot-separated identifiers (e.g., 1.2.feature-x.3) that unify the concepts of linear releases, branches, and drafts into a single, hierarchical structure.

  • Evolution: The process by which an AI-Native system autonomously or semi-autonomously adapts, improves, and changes its own structure, logic, and capabilities over time in response to new data, feedback, or changing goals.

  • Emergent Identity: A recognizable pattern that arises from the web of relationships connecting a collection of Ideas. Identity is not an intrinsic property of a single Idea, but a quality that emerges from its context.

  • idea:: A URI scheme for navigating the versioned, branched reality of the Idea ecosystem. It provides syntax for both simple, dynamic requests and fully resolved, permanent references.

  • Identity Relationships: The set of connections that give rise to an Emergent Identity.

    • Lineage: The historical chain of versions that links an Idea to its past and future iterations.
    • Causality: A relationship where one Idea directly causes or triggers the creation of another.
    • Grouping: The mechanism for bundling multiple related Ideas into a single, manageable unit.

  • State Message: A context message containing a state object and an optional schema. It represents a persistent state maintained between ticks of an agent's execution loop.

  • Data Message: A persistent context message containing a data value and an optional schema. It is retained across an agent's execution loop to provide a stable, structured context.

  • Plan: A context message carrying a data-flow graph of Tool Calls that represents an agent's strategy. It is passed between steps to enable iterative execution and adaptation.

  • Process Idea: A self-contained Idea that captures a strategic Plan and its live execution state. Its schema is the library of Tools, its solution is the new Plan for the current tick, and its context contains the Input, State, and the previous Plan.

  • Vessel Idea: A self-contained Idea that is both the definition of a reactive capability and the record of its chosen reaction. Its schema defines the full universe of possible Tools, and its solution captures the specific Calls (instances of those Tools) that were chosen in response to a stimulus.

  • Instancing: The process of handling multiple, independent Instances (each with its own unique identifier and corresponding State message) within a single agent request.

  • Request: A single, self-contained LLM invocation, which takes a context and a schema and produces a solution.

  • Instance: A single, uniquely identified state object and its associated context, processed as one of many within an Instancing operation.

  • Reactor: A specific Idea Transformer designed as a universal runtime for stateful, turn-based agent interactions. It takes a game/process state (Idea) and produces the next state (Idea).

  • Refiner: A specialized Idea Transformer that evolves an Idea's structure. It takes an Idea and a prompt as input and produces a new Idea with an updated schema and migrated solution.

  • Search Path: An ordered list of branch names that defines the retrieval mechanism. It tells the resolver which partitions to look in, and in what order of priority, creating a cascading overlay system.

  • Sovereignty: The principle that an Idea's identity and canonical location are controlled by its owner through a decentralized mechanism, typically DNS. It establishes ownership over a namespace in which Ideas reside and evolve.

  • Variable Reference: A string with a special syntax (†<kind>.<path>) used in a Tool Call's parameters to reference a value from the context (e.g., from a State or Input message). It allows an agent to dynamically wire tools together by using the output of one tool as the input for another.

  • Solution: The structured object returned by a Request. It contains the proposed Tool Calls and the final output, conforming to a predefined JSON Schema.