The latest Developer tools stories across our sources, prepared from the publishers’ own reporting.
The GitHub Blog published a beginner's guide to creating customizable interfaces called canvases in the GitHub Copilot app using the /create-canvas skill. Users describe a desired workflow in plain English and the agent builds the interface, which can be refined, saved as an extension, and shared with a team.
AWS published a post describing an architecture that combines Amazon EKS, Elastic Fabric Adapter (EFA), and DeepEP to support large-scale MoE reinforcement learning. Across 48 P5en instances running a super-sparse MoE model, enabling DeepEP over EFA increased aggregate RL rollout throughput by 40 percent.
An AWS Machine Learning Blog post walks through running SkyRL, an open-source reinforcement learning framework, on Amazon SageMaker HyperPod to train a Qwen3-VL-8B vision-language model to navigate visual mazes with GRPO. Starting from the VisGym SFT checkpoint, the post reports the maze solve rate rose from 43.75% to more than 95% on a fixed 64-maze evaluation set.
AWS published a walkthrough for deploying the public Qwen3-TTS-12Hz-1.7B-Base text-to-speech model from Amazon SageMaker JumpStart to a fully managed real-time inference endpoint, including voice cloning from a short reference clip and transcript.
AWS outlined how the latest Model Context Protocol specification removes protocol-level sessions, sticky-session requirements, and shared session stores for remote MCP servers, allowing requests to reach any server instance. The change simplifies horizontal scaling while shifting state management, retries, observability, and idempotency concerns to other layers.
Simon Willison released commit-rewriter 0.2, adding support for branches other than the default branch, runnable with uvx commit-rewriter --branch other.
Datasette 1.0a41 has been released, adding OpenTelemetry support contributed by Alec Garcia. The release also refactors all of Datasette's modal dialogs into a single Web Component, now documented for other plugins to use.
LangChain announced LangSmith Engine v2, which adds Red Teaming for proactive agent issue detection and automated validation of proposed prompt and code fixes before human review. The release also expands detection to inefficient agent paths and trends in error rate, latency, and cost.
LangChain launched Trajectories in LangSmith, a chronological, conversational view of an agent session that aggregates messages from humans, AI, and tools across main agents and subagents. It is available now on all plans in the US, and trajectories can be scored with online evaluators and routed to annotation queues or datasets.
LangChain launched Managed Deep Agents 0.8, adding user-owned credentials, user-level memory, HTTP channels, file transfer in Slack, and a pre-built web search tool powered by Parallel. The release is in beta and web search is available free during that period.
LangChain launched LangSmith Fine-Tuning and smithtune, a CLI that turns agent trajectories stored in LangSmith into fine-tuned models, handling dataset creation, training via Fireworks or Baseten, and evaluation. It is available in Public Beta.
NVIDIA published a tutorial on training mixture-of-experts biological foundation models using Transformer Engine primitives, including GroupedLinear and a fused MXFP8 GroupedMLP kernel. A benchmark on eight NVIDIA B200 GPUs reported up to 2.21x the throughput of a Hugging Face baseline.
Simon Willison commented that Cloudflare has shipped support for the Vary header, a change he says he has wanted for years. He described the classic problem of serving HTML or JSON based on the Accept header, which he said was previously impossible to cache safely behind Cloudflare.
NVIDIA Cluster Readiness Engine (NVCRE) is an open source Kubernetes controller that runs real distributed workloads to validate GPU cluster readiness before production workloads. It attributes failures to specific nodes and categories, supports adaptive fault isolation, and integrates with NVIDIA AI Cluster Runtime and NVSentinel.
NVIDIA introduced NodeWright, an open-source, Kubernetes-native package manager that declaratively configures and updates host operating systems across GPU clusters. It orchestrates cordon, drain, apply, interrupt, and uncordon sequences per node and supports progressive fleet rollouts through DeploymentPolicy resources.
Simon Willison published a tool that explains CSS shadow roots using interactive live examples, noting it was built with a prompt to Fable 5.1 Medium.
NVIDIA introduced SWE-Serve, a benchmark of 53 tasks derived from 83 merged SGLang pull requests, developed with input from the SGLang team. On 19 tasks with live-serving checks, the same patches passed 45.9% with the complete verifier versus 69.4% without those checks.
Graphify is an open-source tool launched in April 2026 that converts codebases and unstructured data into queryable multimodal knowledge graphs to support AI coding workflows. Recent updates added parser features and cross-file resolutions, while community feedback cites strong architectural appeal alongside early-tool integration friction.
Google has added a Kernel Profiling suite to XProf, its open-source profiler for TPU workloads, letting developers see cycle-level details in custom Pallas kernels that previously appeared as single opaque blocks in trace captures.
Simon Willison is co-hosting a Birds of a Feather session with Jesse Vincent in San Francisco on Wednesday, 14th October for people building with and on top of coding agents. The evening is described as an informal show-and-tell for early, odd or unfinished experiments rather than product pitches.