
OneCLI Shields AI Agents From Leaking API Keys With a Credential Gateway
OneCLI is an open-source gateway that stores API keys once and injects them into agent requests at runtime, so agents never touch real credentials.
20 stories

OneCLI is an open-source gateway that stores API keys once and injects them into agent requests at runtime, so agents never touch real credentials.

A community quantization recipe shrinks Qwen3.5-9B to roughly 4 GB at under 4 bits per weight while beating standard low-bit GGUFs on benchmarks.

Xiaomi's open MIT-licensed omnimodal MoE handles text, image, video, and audio in one 1M-token model, activating just 15B of 310B params.

Aphrodite Engine is a vLLM-derived inference server that keeps the throughput but adds nearly every quantization format and sampler community models actually ship with.

Xiaomi's new open-source MoE runs 1.02T parameters with a 1M token context, targeting long-horizon coding agents at a fraction of Opus pricing.

Unsloth releases quantized GGUFs of Mistral's new 128B unified model, making a frontier-class reasoning, coding, and vision model runnable on ~64GB RAM

An open-source multi-agent stack orchestrates hypothesis generation, code execution, visualization, and report writing under a LangGraph state machine with human review.

A Go-based MCP server wraps Xiaohongshu with browser automation, letting Claude, Cursor, and other AI clients post, search, and engage on the platform.

An open-source Next.js pipeline that turns raw scripts into full animated videos through character extraction, storyboarding, keyframe generation, and multi-model video synthesis.

Omi bundles a wearable, mobile apps, and a self-hostable Python backend into a single MIT-licensed stack for always-on transcription and AI recall.

TDD Guard hooks into Claude Code to block implementation without failing tests, keeping the coding agent honest across nine test frameworks.

ScrapeGraphAI is an open-source Python library that turns plain-English prompts into structured web data using LLMs and graph-based scraping pipelines.

Einsia released OpenChronicle, an MIT-licensed local-first memory layer that captures your Mac's accessibility tree and feeds any tool-calling agent.
antirez uploads custom GGUF quantizations of DeepSeek V4 Flash to Hugging Face, purpose-built to run the 284B MoE model on MacBooks with 128 GB of RAM

A community-modified Qwen3.6-27B ships with refusals surgically removed, AWQ 4-bit quantization, and native multi-token prediction hitting 110+ tokens per second.

claude-peers-mcp gives every Claude Code session a live message bus, letting parallel AI agents coordinate directly without a human in the loop

Trigger.dev shipped chat.agent, a durable primitive that gives every conversation its own long-running Linux machine that sleeps between turns and survives crashes.

KoboldCpp v1.118.1 adds distributed multi-GPU inference over the network, LTX2.3 video generation, and llama.cpp RPC interoperability.

SeeSee21 released Z-Anime, a full fine-tune of Alibaba's Z-Image on anime data, running on 8GB VRAM with 4-step distilled variants

A 27B-parameter Qwen variant arrives on Apple Silicon with per-layer mixed-precision quantization, beating uniform 4-bit at the same memory budget.