A Logos-grounded theological source architecture for AI-era governance, making assumptions about truth, personhood, authority, delegation, provenance, and human accountability explicit, traceable, and reviewable.
- New human readers: start with this README.
- AI agents and coding assistants: start with
AI_WORK_START_HERE.md. - Repo map and document index: see
AI_TABLE_OF_CONTENTS.md. - Cross-repo data flow: see
DATA_FLOW_MAP.md. - Governance conventions: see
docs/governance/README.md. - Roadmap: see
docs/roadmap/theological-buildout-roadmap.md. - Maintenance runs: see
docs/maintenance.md.
This repository is a theological source architecture for AI-era decision systems.
It is also the upstream governance architecture for the companion
logos-scripture-graph
data-plane repository. That link is a governance_contract: this repo defines
source-trust, derivation, review, and authority meaning; the Scripture Graph repo
implements deterministic Scripture data artifacts under that governance. The
paired GitHub issues are
logos-governance-architecture#54
and logos-scripture-graph#7.
Child Logos repos are implementation/support planes, not higher authority over the governance registry or canonical Scripture. Noesis Atlas may connect only as non-governing, non-mutating advisory comparison context.
It is the theological upstream companion to the AIRCA Fractal Decision Architecture project.
In simple terms:
- AIRCA repository = operational decision architecture
- This repository = theological source architecture, doctrinal taxonomy, ordering logic, weighting logic, derivation chains, and Christian instantiation logic for LAIRCA-style use
The purpose of this repository is to make the theological layer explicit instead of leaving it hidden behind governance language, optimization language, or technical tooling.
It is meant to be useful for human readers, future AI interrogation, structured retrieval, governed machine-readable artifacts, and institutionally serious downstream applications.
No decision architecture is neutral.
Every serious system already assumes answers to questions about reality, truth, personhood, authority, responsibility, delegation, and what counts as a good outcome.
Every architecture already assumes answers to questions like:
- What is a person?
- What is truth?
- What may be delegated?
- What does responsibility require?
- What goods outrank others?
- What kinds of efficiency gains are morally acceptable?
The Logos project makes those assumptions visible, structured, traceable, and reviewable.
Many decision systems pretend to begin at the level of process, policy, data, or optimization. In reality, every serious system already contains deeper assumptions about reality, truth, personhood, authority, responsibility, and what counts as a good outcome.
This repository exists to name those assumptions at the theological level and organize them in a way that can be carried downward into:
- ethics
- governance
- LAIRCA configuration
- institutional decision design
- AI-assisted workflows
- nonprofit and ministry applications
The project therefore starts upstream of process design. It starts with theology.
AI-era institutions face a recurring temptation: to move directly from capability to deployment while skipping the deeper questions that should govern design.
This repository is meant to slow that process down in a productive way.
It asks:
- What kind of anthropology is being assumed?
- What remains accountable to human judgment?
- What kinds of ranking are morally dangerous?
- When is automation appropriate, and when does it become deforming?
- How should dignity, truth, prudence, and stewardship function in architecture rather than slogans?
In that sense, the project is not anti-tool. It is anti-deformation.
No model output can certify its own governance validity.
A model may assist with retrieval, summarization, drafting, classification, ranking, recommendation, or comparison. But validity must come from outside the model through source provenance, authority boundaries, review obligations, derivation chains, audit evidence, and accountable human judgment.
The repository organizes governance as a chain of derivation:
Theological / moral source
-> anthropological claim
-> ethical implication
-> governance requirement
-> design constraint
-> AI workflow rule
-> review obligation
-> audit evidence
The purpose is not to make every decision mechanical.
The purpose is to prevent hidden assumptions from silently becoming automated power.
This repository should be read as a companion, not a replacement, to the AIRCA Fractal Decision Architecture repository.
The AIRCA repository focuses on the operational shape of Architect, Inform, Rank, Commit, and Act. This repository focuses on the theological and metaphysical assumptions that can configure one specifically Christian instantiation of that architecture.
The relationship is best understood as:
Logos theological source architecture
-> LAIRCA Christian instantiation
-> AIRCA operational decision architecture
-> institutional application
That means this repository is upstream of Christian uses of AIRCA while remaining distinct from AIRCA as an operational framework.
The repository preserves a fractal architecture because it is designed to grow through repeated internal structure rather than uncontrolled accumulation of notes.
In practical terms, that means:
- important concepts should become nodes
- the same structural logic should repeat at multiple levels
- parent-child and derived-from relationships should remain explicit
- the architecture should be extensible without redesigning the whole shell every time a new concept is added
A doctrine node, a canon node, a derivation, a weighting profile, a scripture node, a lexical node, and a use-case example should all be capable of fitting into a shared recursive logic.
The shell should not be rewritten every time the project expands. Growth should happen by adding nodes and relationships inside the recurring structure.
A doctrine is not the same thing as the practical sequence in which it is consulted. Stable theological objects should remain distinguishable from interpretive priority stacks.
Ordering explains sequence. Weighting explains force. The repository needs both because theological judgment often fails when those two layers are collapsed into one another.
The project should make it possible to trace how a theological claim becomes an ethical implication, then a governance requirement, then a design constraint, and finally an institutional application.
This repository should remain readable as prose, but also capable of later machine interrogation through explicit metadata, recurring patterns, and clear relationships.
Tool power, model fluency, or operational convenience should not define the architecture. Those belong downstream of theology, anthropology, prudence, and accountability.
The repository should be broad enough for serious Christian reuse while still allowing optional overlays for narrower traditions, local doctrinal commitments, or institution-specific emphases.
The repository should preserve where reality resists current models through a governed exceptions-lake layer.
Expectation, exception, and adaptation should remain distinct objects so learning remains traceable.
See:
docs/governance/exceptions-lake-and-learning-loop.mddocs/governance/exceptions-lake-integration-note.mddocs/roadmap/exceptions-lake-learning-loop-roadmap-extension.mddata/graph/schemes/exceptions-lake.md
This repository depends on disciplined recursive structure and controlled vocabulary.
Because it is designed for human reading, semantic retrieval, RAG-style retrieval and interrogation, recursive comparison, future ontology development, and graph-oriented concordance growth, contributors should avoid casually inventing new anchors, tags, node types, relationship verbs, or category names.
Preferred order of extension:
- model the new thing as a node first
- fit it into existing structure and vocabulary where possible
- expand vocabulary only when the concept cannot fit cleanly
- register new vocabulary before broad use
- avoid structural redesign unless the current shell cannot represent the concept without distortion
Vocabulary drift is not a minor formatting issue here. It weakens retrieval, comparison, ontology coherence, and future reuse.
For detailed governance conventions, see docs/governance/README.md and AI_TABLE_OF_CONTENTS.md.
This repository is not only concerned with what a node says. It is also concerned with where a node comes from.
Because the project is designed to support human reading, semantic retrieval, recursive comparison, and future ontology development, major nodes should eventually preserve provenance more explicitly.
That means the repository should be able to distinguish between:
- material directly derived from one major source
- material synthesized from several Christian thinkers or traditions
- material adapted from earlier categories into a new theological frame
- material appropriated from non-Christian philosophy or adjacent intellectual traditions
- material that requires translation before meaningful comparison can occur
This matters because theological architecture becomes less trustworthy when synthesis is presented as if it came from a single source, or when borrowed philosophical categories are treated as if they were native to Christian theology without acknowledgment.
In practical terms, the project should increasingly support governed distinctions such as:
- direct derivation
- multi-source synthesis
- appropriation
- adaptation
- semantic alignment
- misalignment
- translation burden
That discipline is especially important in cases where Christian thinkers are deeply shaped by non-Christian philosophical traditions while also revising, correcting, or transforming them.
Over time, this repository should support a clean architecture for both:
- Christian canon thinkers
- non-Christian philosophical source traditions
That will allow the project to model intersections such as:
- Augustine and Platonism
- Aquinas and Aristotelianism
- Christian doctrine and modern political or philosophical categories
- areas of strong alignment, partial alignment, misalignment, correction, or required translation
The goal is not to flatten these relationships into simple influence language. The goal is to preserve traceable theological and conceptual lineage.
This project is gradually building:
- a canon and source map
- a doctrine taxonomy
- ordering profiles
- weighting profiles
- derivation chains
- LAIRCA configuration logic
- AIRCA crosswalks
- worked examples
- future machine-readable artifact patterns
- a theological buildout roadmap
- a scripture and interpretation layer
- original-language, translation, and manuscript support layers
- a noncanonical and boundary-source control layer
- graph and concordance surfaces for governed machine-readable relationships
- a future primary-sources corpus layer for witnesses, fragments, transcriptions, passage reconstructions, lexical evidence, and translation comparison
- a future confidence-aware textual knowledge system that could let users move from witness or fragment to passage, to original-language analysis, to translation comparison, to doctrine and theology without losing provenance, confidence, or context
The intent is that the repository becomes not merely a collection of theological notes, but a usable theological architecture.
The repository is being built in this order:
- Foundation
- Canon and source map
- Doctrine taxonomy
- Ordering profiles
- Weighting profiles
- Derivation chains
- LAIRCA layer
- Integrations and crosswalks
- Worked examples
- Research expansion
That order is intentional. It is meant to prevent the project from jumping too quickly into application before the theological source architecture is stable enough to support it.
For the practical sequence of which thinkers, concepts, and comparisons to build next, see docs/roadmap/theological-buildout-roadmap.md.
For repo-wide integration of doctrine, scripture, source-control, and graph layers, see docs/roadmap/repository-integration-map.md.
The repository already contains:
- a roadmap
- canon and doctrine branch foundations
- a full default ordering profile
- a full AI governance weighting profile
- a full derivations guide
- a full LAIRCA guide
- an anthropology doctrine node
- an AIRCA companion crosswalk
- a full worked example deriving anthropology into AI governance workflow
- a growing scripture layer
- a growing original-language layer
- translation and manuscript control layers
- a noncanonical and heresy-boundary layer
- graph and relationship-object entry points
The strongest next steps are:
- build more doctrine nodes
- build canon source nodes
- deepen scripture, lexical, and manuscript linkages
- build additional biblical theme nodes
- add comparative ordering profiles
- add nonprofit and ministry weighting profiles
- add more worked derivation examples
- expand graph and concordance structures only where governed relationship objects are truly needed
- introduce machine-readable sidecars once the prose architecture is stable enough
- continue scaffolding the future primary-sources branch so it can eventually support witness-, fragment-, lexical-, and translation-aware textual analysis
For the recommended theologian sequence, thinker-page checklist, concept-promotion logic, comparison roadmap, and synthesis roadmap, see docs/roadmap/theological-buildout-roadmap.md.
The repository includes a graph-oriented layer for governed machine-readable relationships.
That graph layer should remain:
- downstream of the human-readable theological architecture
- governed by the same vocabulary and trust controls
- aligned with scripture, doctrine, canon, translation, manuscript, and boundary-source rules
See data/graph/README.md.
The repository is also being designed so it can one day grow into a governed biblical primary-sources corpus.
That future layer would not be a flat archive of manuscript images or quotations. It would be a structured and confidence-aware source system for:
- manuscript witnesses, including fuller witnesses such as codices, books, letters, or substantial continuous-text objects
- fragments where a witness survives in pieces
- transcriptions
- passage reconstructions
- lexical evidence
- translation comparison across modern languages
- provenance, confidence, and trust-aware source handling
The long-range goal is to make it possible to move through layers such as:
witness or fragment
-> transcription
-> passage reconstruction
-> lexical evidence
-> translation comparison
-> doctrine / concept / theology
without collapsing those layers into one another.
If built well, this future branch could help a user or AI system:
- inspect early witnesses for a passage
- compare known witnesses across a verse or passage span
- trace an original-language word or phrase in context
- compare likely meanings across biblical and relevant non-biblical usage
- compare renderings into modern languages
- preserve uncertainty, reconstruction limits, and provenance rather than forcing false certainty
This future layer is intended to extend the repository’s existing ontology rather than replace it. It belongs inside the same fractal shell, governed by the same commitments to stable identity, typed relationships, trust zones, provenance, validation, and machine-legible structure.
See:
docs/roadmap/biblical-primary-sources-future-framework.mddocs/primary-sources/README.mddocs/primary-sources/ontology-and-taxonomy.mddata/graph/schemes/primary-sources.md
Operational maintenance commands have been moved out of the public landing page.
See docs/maintenance.md for staged cleanup/buildout runs, checkpoints, logs, reports, and common commands.
Author attribution should appear as Lowell T. Wong.
This repository is licensed under CC BY 4.0.