Zenith Logic Foundry

zlf Usage Guide

zlf is a RocksDB-backed graph database with a single Warren Abstract Machine (WAM) Prolog runtime. It persists graph facts, evaluates Prolog rules and queries, and layers BM25 / vector / temporal / code retrieval on top.

Primary interfaces:

This page is the directory. Each section below links to a focused document.

Guides

Topic Document
Build, configuration, quick start getting-started.md
Code indexing: Tree-sitter ingestion, graph, retrieval, benchmark code-indexing.md
Prolog: facts, rules, queries, terms, builtins prolog/prolog-basics.md
Prolog: nodes / edges / properties / labels and shortcuts prolog/prolog-storage.md
Prolog: logical facts (FactStore), mutation, retraction prolog/prolog-storage.md
Prolog: tabling and stratified negation prolog/prolog-negation-tabling.md
Database: property-graph model database/graph-model.md
Database: BM25 / vector / temporal / code retrieval database/retrieval.md
IndexProfile: selecting fields and derivative indexes index-profile.md
Interface: Prolog REPL interfaces/repl.md
Interface: JSON-over-STDIO interfaces/json-stdio.md
Interface: HTTP server interfaces/http-server.md

Advanced implementation details

Document Topic
internals/wam.md The actual WAM runtime: machine, executor, instruction set, providers
internals/rocksdb-layout.md RocksDB key layout and index structure
internals/index-algorithms.md BM25/Tantivy, HNSW + exact fallback, temporal indexes, tabling

Archived

Outdated or superseded material lives in archive/: