Code and research referencesActive developmentBuilding in public

Built in public. Available to inspect.

Research and architecture in development. This page explains our design direction. It is not a list of completed PRAXIS capabilities. See the documented product capabilities and open work.

Conflict and policy intelligence infrastructure should be transparent and extensible. Start with DIALECTICA v3 for current capsule development. Earlier DIALECTICA work, KAIROS, AGON, the reference pipeline, and the ACO grammar document the research behind it. Read each repository for its status, license, and setup instructions. These references are not a complete self-hosted copy of PRAXIS.

DIALECTICA v3 on GitHub Knowledge Pipeline

Open stack

Open the layers that must be inspectable.

The reference code shows how source records become graph context. Inspect the implementation, development status and license before building on it.

Spec

ACO ontology

The shared grammar: primitives, subclasses, relationship types, graph layers, and serialization rules.

Pipeline

Extraction and validation

Source ingestion, primitive extraction, confidence handling, source-span binding, and schema validation.

Graph

Dialectica engine

Temporal graph construction, contradiction edges, commitment state, graph queries, and context packs.

Products

Professional surfaces

PRAXIS is the analytical workspace. Wind Tunnel and CONCORDIA explore related research questions. Shared architectural goals do not imply completed product integrations.

Choose a repository by purpose and contract.

DIALECTICA v3

Rust/Python · Apache-2.0 · in-progress

Current v3 capsule reference path; use the pinned build guide. Hosted integration and PRAXIS promotion require separate checks.

KAIROS — temporal vision

Rust · Apache-2.0 · historical reference

Historical research; not the canonical v3 capsule contract.

AGON — conflict vision

Historical research; not the canonical v3 capsule contract.

TACITUS Knowledge Pipeline

Python · Apache-2.0 · historical reference

Historical research; not the canonical v3 capsule contract.

Dialectica (A2)

Python/TypeScript · Apache-2.0 · historical reference

Historical research; not the canonical v3 capsule contract.

TACITUS Ontology (ACO)

Python · See repository terms · in-progress

Evolving vocabulary reference; not a complete v3 capsule implementation.

Pinned source, prerequisites and verified fixture commands. Check repository and asset terms separately.

Repositories

Current development. Earlier reference work.

Use DIALECTICA v3 to explore the current local workflow. The other repositories offer research and implementation references for temporal analysis, disagreement, extraction, and ontology design.

DIALECTICA v3 — current development

Under development

sargonxg/A3_DIALECTICAbyTACITUS_v3

The current capsule and knowledge-engine development. Start here for the local compiler, schemas, review records, and reference interfaces described in the build guide.

Open source

Apache-2.0

RustContext CapsulesReviewCLIMCP

What's inside

>Local document-to-capsule reference workflow
>Versioned schemas and validation
>Review decisions and source traceability
>CLI and local MCP reference interfaces
>Check the README for current prerequisites and limitations

View on GitHub

KAIROS — temporal research reference

Historical reference

sargonxg/KAIROS-temporal-vision-TACITUS

A Rust research reference for extracting and organizing dates, events, commitments, and temporal relationships from text. Its ideas inform the temporal side of the knowledge-engine design.

Open source

Apache-2.0

RustTemporal GraphAllen-13GeminiAxum

What's inside

>Deterministic Rust core: dates, source spans, diagnostics, relations, graph assembly
>Pre-read segmentation, tension-marker detection, actor canonicalization
>Allen-13 temporal relations between episodes
>Embedded Axum web workbench with Gemini-backed extraction
>Crisis-dossier demo: 30+ dates, 25+ events, 10+ episodes, 80+ Allen relations

View on GitHub

AGON — conflict research reference

Historical reference

sargonxg/AGON

Source-available Rust research on actors, claims, denials, commitments, and contradictions. All rights are reserved: public visibility does not grant permission to reuse or redistribute the code.

Public research reference

RustPostgresVertex AICloud RunConflict Intelligence

What's inside

>Actor, claim, event, commitment, pattern, and contradiction extraction
>Source-backed evidence spans with verified or unresolved status
>Actor-by-actor friction matrix, escalation signals, review questions
>Persistent typed sessions in Postgres; streaming and non-streaming APIs
>Reference work on validation and quality checks; hosted availability is separate

View on GitHub

TACITUS Knowledge Pipeline

Historical reference

sargonxg/TACITUS-Knowledge-Pipeline-open

An earlier reference pipeline for ingestion, extraction, structuring, and queries using the eight core ontology concepts. Useful for studying the approach; see the build guide for the current capsule tooling.

Open source

Apache-2.0

PythonOntologyKnowledge GraphPipeline

What's inside

>8-primitive Conflict Grammar (Actor, Claim, Interest, Constraint, Leverage, Commitment, Event, Narrative)
>4-stage pipeline: Ingest, Extract, Structure, Query
>Sample conflict datasets for testing and development
>Pydantic v2 schemas for every ontology primitive
>Repository-specific installation instructions and example runs

View on GitHub

DIALECTICA A2 — earlier reference

Historical reference

sargonxg/A2_DIALECTICAbyTACITUS

An earlier DIALECTICA reference exploring ontology-guided extraction and graph-based analysis. The current development direction is DIALECTICA v3; this repository preserves prior research and implementation work.

Open source

Apache-2.0

PythonTypeScriptNeo4jLLMNeurosymbolicOAG

What's inside

>Full Ontology Augmented Generation (OAG) pipeline
>Neurosymbolic architecture: deterministic graph + probabilistic LLM
>Typed knowledge graph with 4 reasoning layers (GND, CTX, EVD, RZN)
>Agentic Conflict Ontology with 41+ classes
>Temporal ordering, causal chains, full provenance tracking

View on GitHub

TACITUS Ontology (ACO)

In Progress

sargonxg/tacitus-ontology

Reference schemas for the Agentic Conflict Ontology, including its core concepts and proposed extensions. Check the repository for supported representations and revision status.

Public research reference

See repository terms

PythonOntologyPydanticOWL/Turtle

What's inside

>Canonical ACO spec (8 primitives, 41+ typed classes)
>Dual serialisation: Pydantic and OWL/Turtle
>RFC process for community revisions
>Subclass extensions for domain-specific ontologies

View on GitHub

Role in the engine

Two Rust vision layers. One neurosymbolic composer.

AGON and KAIROS are the two Rust vision layers we are building in the open. AGON reads conflict structure. KAIROS reads time. DIALECTICA composes both into the typed, temporally honest graph that PRAXIS, Wind Tunnel, and CONCORDIA reason over.

Conflict vision (Rust)

AGON

Reads who, what, and why. Actors, claims, denials, commitments, contradictions, escalation, quality gates. Preserves disagreement as data.

Temporal vision (Rust)

KAIROS

Reads when. Dated events, episodes, Allen-13 relations, commitment trajectories. The temporal scaffolding for everything else.

Neurosymbolic engine

DIALECTICA

The knowledge engine under development. The intended architecture draws on conflict and temporal research through defined interfaces, with review before context is accepted.

Grammar + reference plumbing

ACO + Pipeline

The eight-primitive kernel grammar and the Python reference pipeline that wires it all into a working end-to-end run.

Architecture

Neurosymbolic by design.

The repositories explore a common approach: organize claims and relationships, retain the source evidence, and use language models where they help. They span different stages of the research and are not interchangeable production components.

conceptual_pipeline.py · pseudocode

1# Conceptual pseudocode, not a supported Python SDK

3from dialectica import ConflictEngine, OntologyGraph

4from dialectica.primitives import Actor, Claim, Event

6engine = ConflictEngine(

7 ontology="agentic-conflict-v2",

8 graph_backend="neo4j",

9 layers=["GND", "CTX", "EVD", "RZN"]

10)

12# Ingest -> Extract -> Structure -> Reason

13graph = engine.process(documents)

14insights = engine.reason(graph, query="...")

Conflict Ontology

8 primitives, 41+ typed classes. The schema that makes conflict machine-legible.

4 Graph Layers

GND (recorded facts), CTX (context), EVD (evidence), RZN (reasoning). Separating them helps reviewers inspect an account; it does not establish truth.

Neurosymbolic reasoning

Repeatable queries and language-model assistance serve different purposes. Both depend on the quality of the records and evidence they use.

Source traceability

The design links assertions to source passages and review history. Missing links and unsupported inferences should remain visible.

Contributions welcome

Join us. Build better tools for analysis.

We welcome engineers, practitioners, and researchers who can help test the approach. A useful contribution might be a clearer schema, a difficult source example, an evaluation method, or a better review experience.

Read the license and contribution guidance for the repository you want to work with. Where contributions are accepted, open an issue or propose a change. You can also write to us with an example or research question.

hello@tacitus.me Contribute on GitHub

Engineers

Python, TypeScript, Neo4j, LLMs, graph databases, neurosymbolic AI.

Conflict professionals

Mediators, negotiators, diplomats, HR; domain expertise matters.

Researchers

Ontology engineering, conflict theory, computational social science, NLP.

Ideas and contributions welcome

Get in touch: ideas for new conflict primitives, something built on the pipeline, related research. Everything is in motion. Write at hello@tacitus.me.

Core engine

Dialectica

The neurosymbolic context backend for conflict and policy intelligence.

Learn more

Developer resources

Developers

Open reference artifacts and machine-readable feeds. Contact us to discuss proposed integrations.

Explore

For LLMs

Machine feeds

/llms.txt, /ontology.json, citation guide for agents.

View