2026年8月14日 技术热点总结

项目简介：DeepSeek Harness: Everything is a Plugin.

技术栈：TypeScript、cordis、dsh、dsh-plugin

核心介绍：English | 中文 DeepSeek Harness (dsh) is an open-source agent harness developed by DeepSeek AI. It uses an architecture where everything is a plugin, and is powered by Cordis, whose design is described in _A Programming Paradigm for Spatiotemporal Composability_. DeepSeek Harness is currently in _developer preview_ and is iterating rapidly. THERE WILL BE COMPATIBILITY-BREAKING CHANGES. Install Node.js, then run: npx @deepseek-ai/dsh web

项目数据：⭐ 40,515 Stars，🍴 3,183 Forks

项目简介：Turn any technical book PDF into a Claude Code skill — ready to study, reference, and use while you work.

技术栈：Python、agent、agent-memory、agent-skill、agent-skills、agentic、agentic-ai、agents、ai-agents

核心介绍：book-to-skill Turn any technical book, document folder, or collection of sources into a unified agent skill — ready to study, reference, and use while you work in GitHub Copilot CLI, Amp, or Claude Code. Why · What it generates · Beyond books · How it works · Usage · Install · FAQ · Performance · Changelog 24×–51× fewer tokens than dumping the book into context to answer…

项目简介：Unlocking _everything_ on the CPU with DRAM scrambling

技术栈：C

核心介绍：> *Unlocking everything on the CPU with DRAM scrambling — PSP, C6, microcode, > SMM, and anything else the specs left out.* Poke the DRAM controller and an address can be made to land wherever you want in memory. skitter-creek-bath-salts modifies the bottom layers of the memory hierarchy to rewire the physical DRAM address translations. This scrambles platform memory, exposing protected regions of DRAM — carveouts invisible even

项目数据：⭐ 933 Stars，🍴 94 Forks

技术栈：TypeScript

核心介绍：> ⚠️ No affiliation with any cryptocurrency. OpenMausBot has no token. Any coin using the OpenMausBot, Maus, or SupaMaus name is not created, endorsed, or affiliated with this project or its maintainer. I have received no tokens, payment, or allocation from anyone, and I will not be endorsing any token. An open-source version of Grok Bot — bring-your-own-agent, local-first, on the models you already have.

项目数据：⭐ 827 Stars，🍴 146 Forks

ReRound (Reconstructive Rounding) is a post-training quantization method that addresses the midpoint ambiguity inherent in standard round-to-nearest (RTN) schemes when quantizing weights near the centers of quantization intervals.
Starting from a pretrained LLM, ReRound trains a conditional diffusion model to produce continuous reconstructions of low-bit weights for the LLM. These reconstructed weights act as a guidance signal to disambiguate the rounding direction of weights located close to interval midpoints.
To integrate this reconstruction-guided rounding with conventional RTN, ReR…

Estimating human gaze targets from images in-the-wild is an important and formidable task. Existing approaches primarily employ brittle, multi-stage pipelines that require explicit inputs, like head bounding boxes and human pose, in order to identify the subject of gaze analysis. As a result, detection errors can cascade and lead to failure. Moreover, these prior works lack the flexibility of specifying the gaze analysis task via natural language prompting, an approach which has been shown to have significant benefits in convenience and scalability for other image analysis tasks. To overcom…

The world evolves following its dynamics, i.e., its laws of motion. However, leading video diffusion models largely fit the pixels without modeling how the pixels transit over time. Thus, they render visually plausible frames but may not accurately obey the laws. To capture the dynamics purely from pixels, we introduce Latent Dynamics Reasoning (LDR). LDR casts the latent transition as an explicit kinematic integration, where the lower-order dynamics are integrated numerically and the model regresses only the third- and higher-order residual that drives the rollout. For this integration to …

Large Language Models (LLMs) increasingly act as agents whose procedural knowledge is stored in reusable skill packages and loaded at inference time. As skill libraries grow, a central challenge is to expose the smallest sufficient executable context under a limited context budget. Existing systems struggle to reuse routines below the whole-skill level, preserve procedural contracts during compression, keep compressed routines executable and expandable, and update the compressed library as skills evolve. These challenges reveal a unit mismatch: skills are retrieved as packages, compressed a…

Exploration has been a focus of reinforcement learning research for a long time. Recently, there has been growing evidence that it is also an important ingredient in LLM reinforcement learning recipes that can significantly impact downstream performance. Many existing methods control exploration in the action-space, for example, using temperature scaling. However, these methods cannot reorder tokens but only influence the variance in the output distribution. This limits exploration and can lead to divergence or stalled training. Here, we investigate parameter-space exploration, where rollou…

AI models have achieved remarkable success across diverse domains, yet the mechanisms underlying their capabilities and the risks they may pose remain poorly understood. As AI development becomes faster and increasingly automated, mechanistic exploration remains largely manual, widening the gap between what models can do and our ability to understand and control them. To bridge this gap, we introduce Mechanist, an agentic system that uses AI as a scientific instrument for the autonomous discovery of mechanisms underlying AI intelligence. To support autonomous mechanistic discovery, we const…