semantica-agi/semantica is a self-hostable, open-source context and accountability layer that adds structured intelligence, causal reasoning, provenance, and governance to AI agents with a permanent audit trail. It addresses the problem that AI agents lack accountability and explainability by storing auditable meaning instead of unstructured embeddings.
Project overview
It targets a gap where AI agents store unstructured embeddings rather than auditable meaning, making decisions untraceable for regulators and risk teams, and provides structured memory, decision intelligence, and governance with a permanent audit trail.
Project type
MCP · AI Agent · RAG · Evaluation & Observability
Use cases
Knowledge Q&A · Search & Research · Automation
Deployment
Refer to project documentation
License
MIT
Best for
Enterprise teams that need accountability and explainability for AI agent decisions, including structured provenance and audit trails for regulators and risk teams.
AI engineers and developers building agents that require structured memory with queryable entities, relationships, and facts rather than unstructured embeddings.
Key capabilities
Structured memory layer storing entities, relationships, and facts as queryable graph nodes with typed properties, provenance, and temporal tracking.
Creates first-class queryable objects for AI decisions with causal links, precedent search, and downstream impact analysis.
Swappable storage backends for RDF and Labeled Property Graphs, plus vector stores.
Full-featured MCP server exposing tools for entity extraction, decision recording, and querying.
Limitations and risks
Deterministic reasoning features do not require an LLM; external LLM providers (OpenAI, Anthropic, and others) supported via LiteLLM are optional and needed only for extraction or generative tasks. Teams should account for external LLM costs if those optional features are used.
Getting started
Install with a single pip command (pip install semantica) and import ContextGraph in Python to begin using the structured memory layer. Setup is rated easy.
Evidence and sources
README: Semantica is the Context and Accountability Layer for AI agents: context graphs, decision intelligence, explainable reasoning, and policy governance, with a permanent audit trail…
README: The reasoning engines, KG construction, and provenance layer are fully deterministic; no LLM is required to use them.
Release: v0.5.1
README: pip install semantica
README: delivered **100% open source** and self-hostable, with zero black boxes and zero vendor lock-in