New
ollama/ollama v0.40.5 — Local and cloud model runtime
ollama/ollama is a runtime server that runs open models locally or in the cloud and connects them to a range of tools.
CURATED OPEN SOURCE INTELLIGENCE
New projects
New
surya-koritala/glyd — Lossless Compression of bf16 Weights and KV Cache
A look at the lossless GPU memory compression approach in surya-koritala/glyd and the measured results published in its README.
New
vercel-labs/fx v0.0.14 — A Zig single-binary coding agent
Covers the latest release of the Zig-written coding agent vercel-labs/fx and the limits stated in its README: the documented feature set and its experimental stage.
New
100yenadmin/hermes-teammates v1.0.0 — Named Subagent Teammates
A look at the features and configuration boundaries of 100yenadmin/hermes-teammates, a plugin that calls subagents by name in Hermes Agent and carries execution forward.
New
fesens/opentpu — An Open-Source AI Accelerator Designed by an AI Agent
fesens/opentpu is an open-source AI accelerator designed by an AI agent, gathering everything from hardware to compiler into one repository and presenting run results on a Kintex-7 card.
New
cranebeaker/coop-anything — Procedure for turning single-player games into co-op
cranebeaker/coop-anything is a repository that provides agents with procedures and tools for making single-player games playable with friends.
New
srkrz23/vramkit — two-track VRAM compression for weights and the KV cache
srkrz23/vramkit is a Python repository implementing a lossless weight codec and KV cache compression that applies different quantization axes to keys and values.
New
mksglu/context-mode v1.0.169 — Coding agent context management
mksglu/context-mode is a TypeScript tool that manages coding agent context through tool output sandboxing and session memory.
New
cmu-l3/reuf2f — A Lean Benchmark Built on Undergraduate Research Problems
ReuF2F from cmu-l3/reuf2f is a Lean benchmark that evaluates coding agents on 244 open problems from undergraduate research.
New
l-forster/open-jet v0.5.0 — Offline Agent and CUDA Inference Engine
l-forster/open-jet provides an offline coding agent for local models and a CUDA inference engine for Qwen3.8-27B; the latest release is v0.5.0.
New
jevagentharness/jevagentharness — a model-routing proxy for coding agents
jevagentharness/jevagentharness is a proxy that classifies each step of a coding agent and assigns a model; it covers the README's Terminal-Bench 2.0 measurements and their limits.
New
invarislabs/invaris-agentsec v0.9.0 — agent security testing
invarislabs/invaris-agentsec v0.9.0 is a Python framework that stress-tests security policies in AI agent workflows through adversarial scenarios.
New
tt-a1i/archify v3.1.0 — Interactive Diagram Skill for Coding Agents
tt-a1i/archify is an agent skill that turns ideas and codebases into interactive HTML diagrams.
New
perplexityai/space — a Rust-based agent sandbox
perplexityai/space is a Rust-based sandbox platform that isolates and runs agent and user code.
New
qiz029/dscode v0.7.48 — A coding agent that splits work across sessions
qiz029/dscode is a DeepSeek-based terminal coding agent harness in which multiple sessions check one another and divide up side questions or tasks.
New
indras-mirror/xolotl — a large MoE inference engine for a single GPU
Xolotl, a C++ engine that extends Strata, serves several large models through an OpenAI/Anthropic-compatible server, and its README reports a decode speed of 26.8 tok/s for DeepSeek-V4.1-Flash 552B on an RTX 4090.
New
gastownhall/beads v1.3.1 — a graph issue tracker on top of Dolt
An introduction to how gastownhall/beads manages long-running coding-agent work as a dependency graph, and the limits of what can be verified.
New
storytold/vectorcraft — a pure Rust vector art tool
storytold/vectorcraft is a vector illustration tool written in pure Rust; the latest release is v0.12.0.
New
pipelab-org/pipelock v3.7.0 — AI Agent Traffic Firewall
pipelab-org/pipelock is a Go firewall that inspects agent traffic and produces signed execution receipts.
New
liqiiiii/jeva — Principles and Structure of the System 1 Model
liqiiiii/jeva is a Python repository explaining a System 1 decision model that represents states and candidates independently.
New
cartert27/scion — A Rust reimplementation of the Anthropic design
A look at the structure and self-reported benchmark results of scion, a Rust implementation built from Anthropic's source control redesign article.
New
nandhakishorm/laya v0.4.9 — non-autoregressive decision engine
A look at v0.4.9 of nandhakishorm/laya, the scope of its engine that performs selection, scoring, and yes/no decisions, and the limitations it publicly states.
New
rtk-ai/rtk v1.0.1 — a CLI proxy for cutting token usage
rtk-ai/rtk is a Rust CLI proxy that compresses, routes, and filters development command output to reduce LLM token consumption.
New
storytold/cadcraft v0.7.0 — a Rust-based 2D CAD app
storytold/cadcraft reimplements an AutoCAD-style 2D CAD app in pure Rust, with drawing format support and agent integration.
New
alexsjones/llmfit v1.1.17 — Model recommendations by hardware
alexsjones/llmfit v1.1.17 is a Rust command-line tool that recommends language models suited to your hardware and shares measured benchmarks.
New
nvidia/nemoclaw — Coding Agent Stack for the OpenShell Sandbox
nvidia/nemoclaw is a reference stack that runs coding agents such as Hermes, LangChain Deep Agents, and OpenClaw inside the NVIDIA OpenShell sandbox with managed inference.
New
deusdata/codebase-memory-mcp v0.12.0 — a code graph MCP server
deusdata/codebase-memory-mcp is a local MCP server that provides a code knowledge graph and CBM Lens.
New
unclecode/crawl4ai v0.9.5 — the crawler that turns the web into Markdown
unclecode/crawl4ai v0.9.5 is an open-source crawler/scraper that converts websites into Markdown for LLMs.
New
sno-ai/sno-station v1.3.9 — agent squad harness
sno-ai/sno-station is a locally running harness that lets multiple coding agents share an encrypted memory and hand work back and forth.
New
antirez/ds4 — a local inference engine for Metal, CUDA, ROCm
antirez/ds4 is a C-based local inference engine that runs a set of models on Metal, CUDA, and ROCm.
New
thedotmack/claude-mem v13.39.0 — Persistent Memory Across Agent Sessions
thedotmack/claude-mem is a persistent memory system that compresses records of agent sessions and provides context to later sessions.
New
google-gemini/gemini-cli v0.64.0 — A Gemini agent for the terminal
google-gemini/gemini-cli is an open-source agent that lets you use Gemini in the terminal and connect it to multiple tools.
New
graphify-labs/graphify v0.9.91 — a knowledge graph linking code and materials
graphify-labs/graphify is a Python tool that turns code and related materials into a queryable knowledge graph.
New
nousresearch/hermes-official-plugins — Official Hermes Agent Extensions
The official Hermes Agent plugin repository maintained by Nous Research, which separates per-product integrations out of the core.
New
magnitudedev/magnitude — a local inference engine tailored to the hardware
magnitudedev/magnitude is a Rust-based open-source engine that tunes kernels to the hardware and uses them for model inference.
New
getzep/graphiti — a knowledge graph that records facts as they change over time
Covers getzep/graphiti, which manages agent memory as a graph by tracking how facts change and where they come from.
New
infiniflow/ragflow v1.0.0-rc1 — a RAG·agent combined engine
infiniflow/ragflow is an open-source engine that combines RAG and agent capabilities to provide a context layer for LLMs.
New
opensec-git/pi-bolt bolt-v0.8.7 — AOT Native Build of Pi
opensec-git/pi-bolt is an independent fork that ships the coding agent Pi as a native executable and presents its own performance comparisons.
New
zihan-su/self_gradient_forcing_plus — SGF+ Video Generation
An SGF+ repository that separates context history and denoising parameters in autoregressive video generation, producing rollouts of up to 24 hours of video from a 5-second training window.
New
nullmoth/nvidia-macos-driver v2.0.2 — a Metal driver for NVIDIA cards
Covers what the driver is made of and how far it has been verified: running NVIDIA GeForce cards through Metal on macOS 15 Sequoia.
New
tigerless-labs/agent-memory — a Markdown-based long-term memory runtime for agents
Describes the Markdown storage approach and long-term agent memory retrieval structure of tigerless-labs/agent-memory.
New
flashml-org/freetoken v0.1.3 — a MoE serving engine for desktops
An introduction to what flashml-org/freetoken offers for serving large MoE models on edge hardware, and to the limits of its performance information.
New
bherbruck/solidcraft v0.12.0 — Parametric 3D CAD in Pure Rust
bherbruck/solidcraft is a parametric 3D CAD application that supports desktop, browser and MCP.
No matching stories.