
Executor Collapses 278,800 Tokens of AI Tools Into 1,044 With One Gateway
A new open-source MCP gateway lets one catalog of tools serve every agent, cutting a 278k-token prompt down to about 1k tokens.
20 stories

A new open-source MCP gateway lets one catalog of tools serve every agent, cutting a 278k-token prompt down to about 1k tokens.

A community fine-tune strips refusals from Qwen3.6-27B, then republishes the model as calibrated GGUF quants that hold 94 percent of full-precision accuracy at Q4.

An independent researcher cracked Google's SynthID image watermark using only FFT and a Stable Diffusion VAE, bypassing the detector with visually lossless output

Executor turns any OpenAPI, GraphQL, or MCP server into a single sandboxed tool call, cutting agent context bloat from 278k to 1k tokens.

Agenta bundles a prompt playground, versioning, evaluation harness, and OpenTelemetry-based tracing into one open-source LLMOps stack you can self-host.

An open-source clipping pipeline uses LLM highlight detection, Whisper, and auto vertical cropping to turn long videos into ranked 9:16 shorts.

Splink is a Python library from the UK Ministry of Justice that deduplicates millions of records without unique IDs, running on a laptop.

A Go middleware that wraps DeepSeek's web chat as OpenAI, Claude and Gemini compatible APIs, with account pooling and tool calling built in.

A 20-line grammar file cuts Qwen3.6 thinking tokens by 22x on HumanEval+ with zero fine-tuning, and even lifts pass@1 on LiveCodeBench.

Taskmaster AI turns a PRD into a dependency-aware task graph your coding agent can actually follow, cutting vibe-coding drift at scale

Tencent shrunk a 1.8B translation model to 440MB using ternary quantization, and it outperforms 72B open models plus Google Translate on-device.

Laravel's official MCP server gives AI coding agents deep context about your app, with 15+ specialized tools and 17,000 pieces of framework knowledge.

A single Windows GUI wraps Waifu2x, Real-ESRGAN, Real-CUGAN, RTX VSR, RIFE and more for image, GIF and video super-resolution.

A fully open-source Chrome extension runs Gemma 4 E2B entirely on your GPU — no API keys, no servers, with tab control and semantic history search built in.

The 3D data processing library ships GPU-accelerated raycasting, cross-platform SYCL support, new geometry metrics, and pretrained ML models for point clouds.

An open-source toolkit chains speech recognition, LLM translation, and voice cloning into a one-click video dubbing pipeline with 30+ model backends.

UsefulSoftwareCo open-sourced Executor, an MCP gateway that collapses thousands of API tools into one sandboxed endpoint any agent can call.

An open-source MCP gateway lets any agent share one governed tool catalog, collapsing 1,600+ tools into a single sandboxed call.

OpenPipe open-sourced ART, a framework that trains multi-step LLM agents with GRPO plus an LLM-judge reward system called RULER.

A Claude Code Skill turns 15+ content types into podcasts, PPTs, mind maps, and quizzes by piping everything through Google NotebookLM.