Releases

Local fine-tuning (SFT + LoRA) for Blazen

Local LLM backend for Blazen using mistral.rs inference engine

CLI tool for UCM impact analysis on codebases

Observability and tracing hooks for UCM event streams

HTTP API server for UCM graph queries and event ingestion

Bayesian impact analysis and intent reasoning for UCM graphs

Local LLM backend for Blazen using candle inference engine

OSS Jammi server binary: Arrow Flight SQL, SessionService, TriggerService, and health/metrics endpoints

Data ingestion adapters for UCM (Git, Jira, Linear, code parsing)

Event sourcing and projection layer for UCM graphs

Local image generation backend for Blazen using diffusion-rs (pure Rust Stable Diffusion)

Embedding backend facade for Blazen: picks fastembed (ORT) or tract (pure Rust) per target

Core graph types and entity model for the UCM impact-analysis engine

Cross-platform GPU / CPU / RAM probe for Blazen workers — sources real AdmissionSnapshot fields

Chat SDK for Rust

Headless automation for agentic software engineering workflows

Command-line interface for Jammi AI

Embeddable AI engine for inference, embeddings, vector search, and fine-tuning

Vulkan bindings for Rust

File-system tool suite (read/write/list) for the lash agent runtime.

Vector database, SQL federation, mutable companion tables, and trigger broker for Jammi AI

offline-chess-puzzles - GUI to solve puzzles from the lichess puzzle database

Pure-Rust ONNX inference backend for Blazen embeddings (tract-onnx)

Local embedding backend for Blazen using HuggingFace candle

Candle-native BERT-family encoders for sentence embeddings, with built-in PEFT support via jammi-lora.

Static-dispatch PEFT-style LoRA primitives for candle. Closed-enum dispatch, immutable construction, candle-0.9.x-native.

Shared model download and cache layer for Blazen local-inference backends

Derive and attribute macros for the Blazen workflow engine

JIT (Cranelift) and AOT compiler backend for clojurust

clojurust CLI — run, repl, compile, and eval