Latest Hotfixes · Page 3
4 GitHub repos for agent desktops, finance and testing
Cua for computer use, financial-services for finance documents, Anatomy for diagrams and e2e for plain-sentence tests, read by who confirms the agent is done.
Extend Claude Code with mods, from building to vetting
Whether a mod hook calls next at one of eight gates changes what it does. Call chain, guard, samples, building a mod, and vetting before install.
Microsoft Learn MCP lets AI read Azure docs, no key or fee
We checked six claims about the Microsoft Learn MCP server on its public endpoint: Q&A posts mixed in, a repeated page, budget steps, old class samples.
Read-only Fabric IQ MCP lets AI agents query Power BI
Fabric IQ MCP queries Power BI models read-only with the signed-in user's permissions. See where Power BI security applies and where results go next.
Meta's first frontier model Wonder keeps Full off the API
Wonder Full beats three rivals on 7 of 14 model card tests but isn't available through the API, where Wonder costs $4 in and $20 out and leads on 4.
23K-star text-to-cad lets AI write CAD code for 3D prints
Follow one loop with text-to-cad: build a thin plate as STEP, export STL, measure its FDM wall thickness and fix it. Covers setup, stats opt-out, exit codes.
Google ships 33 Android agent skills, from R8 to Compose
Google's android/skills repo has 33 SKILL.md files. We sort them by folder shape and area, open the AGP 9, R8 and edge-to-edge skills, and cover installation.
Bonsai 2 squeezes a 54GB 27B model into 6GB for laptops
Bonsai 2 27B shrinks 54GB to 5.95GB. We trace the files, matrices, 128-slot groups, rotation, speed and memory, then 14 benchmarks to see where scores dropped.
Jev only judges at $0.042 per 1M input tokens, output free
Who measured Jev's price, multiples, latency and outside tests, and how. Plus rival models, how confidence is computed, and the weaknesses and security report.
llmfit rates which models run on your laptop in one go
Follow llmfit on a 16GB laptop from install to bench: how verdicts are calculated, trust levels for speed numbers, pre-purchase what-ifs and common snags.