Artificial Intelligence

Core playing-card types and abstractions

Tekcesta AI Platform - Privacy-first AI appliances for regulated industries

Tekcesta AI Platform - Privacy-first AI appliances for regulated industries

Decapod is a Rust-built governance runtime for AI agents: repo-native state, enforced workflow, proof gates, safe coordination.

Deterministic graph engine for grounded AI - every query returns fact, inference, or unknown. Never invents.

A Rust client for the new CivitAI site API.

Core data models and traits for Canon Protocol — verifiable, censorship-resistant knowledge sharing for AI

Local embedding generation using GTE-Qwen2-1.5B-instruct via Candle — private, on-device AI inference

Canon node ingestion pipeline — local-first AI knowledge management with anonymous sync

Permanent storage layer for Canon Protocol — censorship-resistant knowledge distribution via Arweave

只需要一个toml即可接入的，支持主流SSG的，声明式的，可持续更新/构建的，全自动的，AI文档翻译智能体。

LLM extraction pipeline for agent memory — Graphiti's pipeline ported to Rust on grit

Embedded, bi-temporal property graph for agent memory: one SQLite file, in-process, deterministic

Graphiti/FalkorDB-dialect query-template translator for grit (pure, fail-loud, bounded inventory)

Our mission is to be the default way developers create custom LLMs

Run prediction functions in your Rust apps.

Low-level, generic LLM provider abstraction for Rust

Gemini provider for anyllm

Cloudflare Workers AI provider for anyllm using the workers-rs binding

Reusable OpenAI-compatible transport and normalization toolkit for anyllm providers

Anthropic provider for anyllm

OpenAI provider for anyllm

Portable unfused ModelPlan reference executor for memra correctness gates

Host-side sampler chain (temperature/top-k/top-p/penalties, llama.cpp CPU sampler semantics) for the memra inference engine

KV-cache format policy (q8_0/q5_1/q4_0/fp8 block layouts) for the memra inference engine

GGUF and HF-safetensors model loading, quant-format decode, and tensor inventory for the memra inference engine

GGUF-native BPE/SPM tokenizer with chat-template support for the memra inference engine

Numeric validation helpers (reference comparisons, tolerance gates) for the memra inference engine

Persistent, git-friendly memory storage for AI agents with hybrid semantic + full-text search

CUDA runtime scaffolding (context, GEMM checks) for the memra inference engine

From-scratch CUDA LLM inference engine for NVIDIA RTX 50-series (sm_120a) and Hopper (sm_90a) - custom kernels, no frameworks

Serving-lane types, SLO admission policy, and engine-truth step stats for the memra inference engine

High-performance, local AI memory with AAAK v3.2 protocol and temporal Knowledge Graph

AI enrichment layer for the mapit codebase mapper — summarization, flaw detection, simulation

Oxen server is a fast data version control backend, supporting local disk and S3. Self host your repositories on your own storage, or use the hosted platform on Oxen.ai. Stores, syncs, and serves versioned datasets, model checkpoints, game assets, studio media, and any large data. Use the oxen CLI to push and pull from the oxen server.

Oxen is a fast data version control system, built with machine learning training data in mind. Designed to handle terabytes of data with ease, using a workflow similar to git. Version both structured and unstructured data of any modality: text, images, video, audio, CSV, Parquet, JSONL, model checkpoints, and more. liboxen is the embeddable core library behind the oxen CLI and server, which power fine tuning and inference pipelines for multimodal LLMs, image models, and video models on Oxen.ai.

The oxen CLI lets you add, commit, push, and diff terabytes of data with a git style interface. Version machine learning training data, model weights, game assets, and studio media. Diff CSV and Parquet files row by row. Self host the oxen server, or push data to Oxen.ai to collaborate on datasets, fine tune models, and run inference. Installs the oxen binary.

A generic, composable genetic algorithm framework for Rust

Reverse engineering assistant that uses a locally running LLM to aid with pseudocode analysis.

Train logisitic regression models for benign vs. malicious files based on byte n-grams and publish research, plus related tools.

IMECE v0.2.0 - Local-First Autonomous Agent Framework. Zero API dependency. Zero cloud lock-in. Full sovereignty.

MCP over a local loopback HTTP endpoint: the transport a desktop AI client uses to reach an already-running GUI instance.

The AI port for Origin: inference the application performs itself, behind a swappable adapter.

The MCP boundary: which of an application's operations an external AI may invoke, and under which permission.

MCP over stdio: the transport a desktop AI client uses to start an application headless.

Trap malicious AI web scrapers in an endless pit of poisoned training data.

plugmem bitemporal memory engine: facts, indexes (BM25, graph, time, vectors incl. HNSW), hybrid recall, snapshot/journal.

A thin CLI wrapper around Docker Sandboxes (sbx) and git worktrees

NCP reference runtime — composable, auditable WASM agent graphs

Pluggable 3D memory extension for Ambi: KV state, semantic vector recall, and rolling summary.