Procedura is world-anchored procedural intelligence built on persistent memory, structured generation, and inspectable artifacts.
Its governing rule is simple: nothing should enter the core unless it deepens continuity between place, memory, action, and world truth.
Procedura is not just a framework, not just a simulation stack, and not just an agent system. It is an attempt to give computation a durable ontology: a way for intelligence to inhabit a world, persist through time, and remain legible through the traces it leaves behind.
The core unit of Procedura is not the prompt, the file, or the API call. It is the anchored entity acting within a world that remembers.
Procedura is a system for building intelligence that is situated rather than detached. It exists to make computation inhabit place, continuity, and consequence.
Its intelligence is not meant to float above the environment. It is meant to arise through interaction with lattice, memory, route, schedule, artifact, projection, and world state.
If a thing works but weakens world-truth, it is foreign to the core.
No meaningful agent action should exist without relation to anchor, place, state, or world context. A native Procedura computation should be answerable in terms of where it is, what it can see, what it remembers, what it is trying to do, and what changed because it acted.
The world should remember through accumulation, not replacement. Authoritative reality should prefer traceable succession over a single mutable present. This protects auditability, causality, replay, branching, continuity, and structural trust.
Interfaces, overlays, and tools are interpretations of world state, not replacements for it. The shell may reveal and orchestrate interaction, but it must not become the ontological center.
Generative capability is real, but native generation should pass through scaffolds, schemas, artifact forms, route structures, and world-aware transforms. Structure exists to preserve legibility, continuity, and survivability under iteration.
Plans, routes, schedules, projections, logs, narrative fragments, topology extractions, and scaffold writes are not byproducts. They are the visible body of cognition. The system should leave behind durable traces.
Abstractions are allowed only if they make the world more intelligible, operable, or generative. If an abstraction only makes implementation easier while obscuring the world, it is suspect.
Procedura should behave like a civilization of modules, not a pile of features. Subsystems must imply and reinforce one another until they feel adjacent by necessity, not by integration work alone.
The system’s identity is not what it says about itself. It is what remains true across implementations: world-anchored intelligence, persistent memory, inspectable artifacts, structured generation, and continuity through place and consequence.
Growth is only valid if the system remains itself at larger size. A feature that works in a prototype but corrupts the grammar of the system at scale is not a real success.
A native Procedura behavior should be reconstructible. A serious operator should be able to understand why it happened, what inputs mattered, what artifact shaped the result, what changed, and where the result lives now.
Procedura is building the systems, structures, and laws required to make that possible.