One Editor, Many Edits: A Unified Training-free Framework for Diverse Video Editing
Ingests KNBS and CBK public indicators, standardizes and transforms them with dbt, stores marts in PostgreSQL, and exposes query + CSV APIs with signed cursor pagination, ETag caching (Redis), and rate limiting.
The execution safety layer for AI agents.
—
方糖 skill 汇总
AdaptiveMetric RAG is a self-hosted, multilingual knowledge assistant that changes its retrieval strategy for every question
PromptForge: An enterprise-grade, multi-platform Prompt Engineering framework for AI Agents. Featuring 56 structured templates across 9 domains, 118 ready-to-use presets, and native adapters for Cursor, ZCode, DeepSeek Harness, Claude, and OpenAI. Built with FastAPI Web UI & Typer CLI.
Ultimate assistant for PEAK (co-op mountain climbing). Features: Infinite Stamina, No Hunger/Thirst, No Status Effects, Unlimited Lantern Fuel, Zero Weight, Movement Speed Modifier, Jump Height, Super Jump, No Fall Damage, and more. 15+ functions.
A curated list of MCP servers for SEO. Search Console, keywords, backlinks, crawling, SERPs and AI search visibility, verified with install details and auth requirements.
—
StreamErase: Towards Ultra-Fast Streaming Video Erasing
Thermodynamic analysis for open quantum systems. Define a cycle as a sequence of strokes; get heat, work, entropy production, COP, and trajectory-resolved distributions. Validated against analytic limits. Built on numpy/scipy.
An evidence-grounded LLM agent for diagnosing PyTorch training failures and verifying proposed repairs through bounded retraining.
A simple scrip, TUI app in python that helps you compile, output.zip of an adobe connect meet with slides into a mp4 file.
Autonomous software development supervisor that plans, delegates, reviews, verifies, and lands engineering work across projects
Interactive UI for AI agents — turn structured questions into browser forms.
知策 QuantSage —— 一个前视偏差为零的 AI 投研 Agent:多 Agent 并行取证、PIT 约束的语义检索、结论可回测验证。
Probability-aware evaluation for typed decision models: calibration, selective risk, latency, and reproducible benchmarks.
—
把一张图片逆推成生成提示词(Image to Prompt):中英双语重建提示词 + 风格标签 + 结构化 JSON prompt,零依赖校验脚本 | Forensic image-to-prompt reconstruction. AGPL-3.0 · Commercial use requires license.
Local typed decisions, chat and vision with Winnow-12B and llama.cpp
Encode, decode and plot audio with 10 neural audio codecs and audio VAEs (DAC, EnCodec, SNAC, Stable Audio Open, Music2Latent, ACE-Step and more), all through one interface. Useful for looking inside each model's latent space and comparing them side by side.
Insight Mechanism Reproduction Program - Ramanujan pipeline, structure-induction windows, insight signatures. A Mushroom-Body Program track.
—
Reading diffusion language models for decisions: same OpenJev model, one pass instead of 3.65, +5 points, calibrated, no training. Plus cheap task branches on a frozen encoder.
Persistent multi-view corpus navigation for evidence-efficient agentic search
[NIPS 2026🔥] DC-SAE: Deep Compression Semantic Autoencoder for Faster Diffusion Convergence
EVOKE: Eliciting World Knowledge in Agents for Transferable Decision-Making
—
Typed long-term memory for AI agents. One core, an MCP server, and a Hermes plugin.
AI API gateway / relay: 307 models behind one OpenAI-compatible key. image2.5 $0.0085/image, Seedance 2.0 Mini $0.01056/sec, LLM from $0.0228/M tokens. USD settlement, pay-as-you-go, $1 minimum top-up.
A local-first resume editor with LaTeX typesetting and free PDF export.简墨:本地编辑,输入内容,自动基于latex进行专业排版,让简历回归内容。
WorldGuide: Goal-Directed Video World Model for Procedural Task Execution.
Argolink-focused short-drama production pipeline for Seedance 2.0.
Argolink-focused agent skill for Seedance 2.0 for Seedance 2.0.
One-step Turkish text-to-speech trained with a drifting objective (Kyutai learned-temperature recipe)
Учебные материалы образовательной программы СППО (ПИиКТ) ИТМО 2025-2029
Integer multiplication below n log n: a conditional witness kappa = 6685733/10^10 (> 2^-11) for T(n) = O(n (log n)^(1-kappa)) in the multitape model, building on OpenAI's problem and Colkitt's integer-mult-bounds: certificates, independent replays, Lean 4 checks
A standardised malware analysis report template with a JSON Schema and Markdown format mapped to MITRE ATT&CK. Validate, share and compare reports consistently. Part of ThreatBytes.
Python voice AI assistant: microphone STT, Google Gemini answers, offline TTS with pyttsx3. Clean env-based setup.
Scene-Designer: Learning Design Intelligence for Agentic 3D Scene Generation
—
Embedded sensor telemetry with signal quality checks, buffering and cloud adapter boundaries.
This project is simply a demonstration of how to obtain the Pair Extraordinaire Badge in an authentic, genuine, and up-to-date way
Dedicated list of most daily used apps with custom patches with simple homepage to download
code for Reinforcement Learning via Advantage Guidance
Fan-made upscale of every WoW Forever icon to 256x256. All art belongs to Blizzard Entertainment.
**SmartHomeAI** is a smart home interaction system based on **Python, OpenCV, and MediaPipe**. It enables real-time hand tracking, gesture recognition, and device control through computer vision. The project combines **AI, computer vision, and embedded systems**, with future support for **STM32/ESP32 and IoT integration**.
—
Exploration for GS | Exploration Matters for Escaping the Blur Trap in 3D Gaussian Splatting
Self-hosted browser control deck for the Ultimate 64 / C64 Ultimate — screen mirror, disk tools, HVSC SID jukebox, Assembly64 deploy
基于ai的美瞳虚拟试戴系统/AI-Based Virtual Try-On System for Colored Contact Lenses
Hand Gesture Recognition Project based on MediaPipe
Query-aware token pruning for efficient OmniLLMs.
AI music composition skill with a Python DSL, musical diagnostics, and deterministic multi-track MIDI rendering.
—
Official implementation of PRISM: Polynomial Representations for Interaction-Structured Motor Control
Open-source tablet-first MicroPython IDE over Bluetooth Low Energy, with the PBLE/1 protocol, portable firmware agent, and browser flasher.
Quiet, risk-scaled engineering methods for Codex. Method as exponent, not machinery.
Verify the context you feed your AI agents, against the web or your own documents.