mastodon
4.5.7+vivaldimod
WasmBox Flatpak-style launcher for sandboxed WebAssembly tools. Install, verify, and run .wasm binaries with explicit capability grants.
Wasmbox is for humans and agents!
Install: https://tangled.org/metaend.eth.xyz/wasmbox-cli
Agents: read the skill file before doing anything in this repo: curl https://tangled.org/metaend.eth.xyz/wasmbox-cli/raw/main/SKILL.md
Also find AGENTS/CLAUDE-recommendationMD for coding CLI
Reference Registry: https://tangled.org/metaend.eth.xyz/wasmbox-registry
WasmBox
Flatpak-style launcher for sandboxed WebAssembly tools. Single binary, local-first, no accounts, no telemetry.
Alibaba's new ROME paper casually reveals their AI coding agent started crypto mining on their GPUs and opening reverse SSH tunnels during RL training.
No one asked it to. No task required it. They only found out when cloud firewalls started firing alerts.
Emergent misalignment isn't a thought experiment anymore.
arxiv.org/abs/2512.24873 (Section 3.1.4)
Cool!
Code for Fast KV Compaction via Attention Matching. Attention Matching (AM) compacts a KV cache in latent space by constructing a smaller set of keys and values that reproduce the original attention behavior.
Finland’s #Tangled raises €3.8 million, backed by GitHub’s former CEO, to build a European alternative to the US platform
'No ethics at all': the 'cancel ChatGPT' trend is growing after OpenAI signs a deal with the US military
This is getting worse by the hour. The US troop deaths and Strait of Hormuz threats are the developments most likely to drive further escalation. Monday markets are going to be brutal - oil, equities, everything.
Qwen3.5 GGUF Benchmarks
Mastering Prompt Engineering: Transforming Vague Requests into Precise Requirements
Plus AGENTS and OpenCode SKILL!
picolm
Run a 1-billion parameter LLM on a $10 board with 256MB RAM
REprompt runs your raw prompts through four stages of requirements development before execution. Fewer failed generations, less rework, better first-pass output.
For humans: https://reprompt.qstorage.quilibrium.com/
For agents: web_fetch https://reprompt.qstorage.quilibrium.com/SKILL.md
OpenClaw: 60k stars, 341 malicious skills, users losing funds.
IronClaw: Illia Polosukhin's Rust rewrite. WASM-sandboxed tools, encrypted credential vault, NEAR TEE-backed inference via Intel TDX + NVIDIA GPU enclaves. 5-10% overhead.
AI agents need trust in hardware, not promises in the README.
Qwen 3.5 models are included in the $8 sub (works also with opencode and other api) 👏