// edition · 2026-05-01
29 stories on AI dev tools, agents, and the coding stack — curated from the day's RSS haul by Agentic Dev's pipeline.
A developer guide warns that installing MCP servers via npx runs arbitrary npm packages with full user-level access, citing incidents including a backdoored postmark-mcp package and CVE-2025-49596, a design flaw affecting ~200,000 instances across Anthropic's official MCP SDKs. The guide proposes...
Anthropic released Claude Security, a codebase vulnerability scanning tool within Claude Code, from closed preview to public beta for Enterprise customers on Thursday. The tool uses multiple parallel agents to analyze data flows and includes a self-validation pipeline to reduce false positives, w...
A benchmark of 13 LLMs on an identical agentic coding task found Claude models via the Anthropic SDK produced 196–203 structured requirements, while models using the OpenAI-compatible SDK produced 13–60, regardless of model size or vendor. The author attributes the gap to scaffolding built into t...
A Dev.to post outlines 10 recommended rules for CLAUDE.md configuration files in Python repositories, covering practices such as mandatory type hints, lockfile management with `uv`, and structured error handling to guide AI coding tools like Claude Code toward project-specific conventions.
OpenAI released Codex CLI version 0.128.0, adding a `/goal` command that causes the coding agent to loop repeatedly until it determines a user-set goal has been completed or a configured token budget is exhausted. The feature is implemented via two prompt templates injected automatically at the e...
The New Stack published a nine-step technical guide for deploying AI systems to production, covering tool interface design, vector search with BM25 reranking, timeout and retry handling, OpenTelemetry-based observability, and bounded agent execution under concurrent load.
Anthropic's Applied AI Team published a prompt engineering guide for Claude, outlining a five-element prompt structure and recommending XML tags as delimiters for organizing information. The guide uses a Swedish car insurance form analysis as a case study, iterating through five prompt versions t...
A developer built HISDashboard, a hospital management AI system using 10 specialized agents distributed across 4 LLM providers with automatic fallback, after a single-provider setup failed due to rate limiting. The system uses a router-specialist-reflection architecture with structured intent cla...
A developer built Claude Skills scripts called "tomorrow" and "today" that aggregate Google Calendar, email, Slack, and iMessage data into prioritized daily briefings, replacing manual checks across six separate apps.
A developer proposed a five-layer governance framework for AI coding agents, arguing that CLAUDE.md alone provides only project orientation, not policy enforcement. The framework adds CONSTITUTION.md, DIRECTIVES.md, SECURITY.md, and AGENTS.md documents alongside runtime enforcement and external v...
A Dev.to tutorial outlines steps for building a multimodal chatbot using Anthropic's Claude 3.5 Sonnet API and React 19, with Vite as the build tool and react-dropzone for image input handling. The guide covers API client setup, environment variable configuration, and component construction using...
GitHub Copilot CLI offers two operating modes: interactive mode, the default chat-like session launched with `copilot`, and non-interactive mode, triggered with `copilot -p` for single-prompt responses without entering a full session.
OpenClaw, an AI agent platform for developer automation, supports Google's Gemini models as an LLM backend, with setup requiring an API key from Google AI Studio. The integration enables agents to process large codebases, multimodal artifacts such as images and video, and supports a free developm...
Arbiter Briefs added financial PDF ingestion to its V2, using regex and heuristics rather than ML to extract metrics from P&L statements, balance sheets, and cap tables. The pipeline uses pdf-parse for text extraction, multer for uploads capped at 10MB and 5 files per analysis, Railway persistent...
A developer used exported AI chat logs, organized in folders by project, to document their workflow for a manager, presenting 12 saved conversations from the past quarter as a record of project decisions and problem-solving approaches.
Anthropic released Claude Opus 4.7 on April 16, scoring 64.3% on SWE-bench Pro (+10.9 points over 4.6) and 70% on CursorBench (+12 points), with added image support up to 3.75MP, a beta token-budget parameter for agent loops, and a new "xhigh" reasoning tier. The company also launched a Managed A...
xAI's Grok 4.3 is now available on Vercel's AI Gateway, accessible via the AI SDK using the model identifier `xai/grok-4.3`. The model features a 1M token context window and a December 2025 knowledge cutoff.
Meta has shifted development focus from its open-source Llama models to a new proprietary model called Muse Spark, built from scratch by Meta Superintelligence Labs. Existing Llama models will remain available, but the company has not committed to continued frontier development of the open-source...
A company migrated its customer support chatbot from Claude 3.5 Sonnet to GPT-5 after a 4-week model evaluation, reporting chatbot CSAT rising from 72% to 92% and tier-1 query resolution improving from 70% to 88% within 30 days of full rollout. Non-English CSAT increased from 61% to 84%, and huma...
Anthropic announced nine connectors on April 28 integrating Claude into creative software including Adobe Creative Cloud (50+ tools), Blender, Autodesk Fusion, and Ableton, built on the Model Context Protocol. The Adobe connector spans Photoshop, Illustrator, Premiere Pro, Lightroom, and Firefly;...
Stripe introduced Link, a digital wallet that allows users to connect payment methods and authorize AI agents to make purchases through approval flows. The product supports both human users and autonomous AI agents spending on their behalf.
GitHub's reliability dropped to 86% this month, down from 90% last month, according to third-party tracking, including a data integrity incident. GitHub leadership attributed the degradation to a 3.5x increase in service load driven by AI usage.
Microsoft launched Agent 365 on May 1, 2026, an enterprise control layer for autonomous agents across Microsoft 365 apps including Word, Excel, Outlook, and Teams. The user-facing component, Copilot Cowork, runs multi-step autonomous tasks and was built in collaboration with Anthropic using Claud...
The UK AI Security Institute evaluated OpenAI's GPT-5.5 for cyber capabilities, finding its ability to identify security vulnerabilities comparable to Anthropic's Claude Mythos. Unlike Mythos, GPT-5.5 is currently generally available.
OpenAI is bringing its models, coding tools, and agentic capabilities to Amazon Web Services' Bedrock platform following a restructuring of its partnership with Microsoft, which had invested a reported $13 billion in OpenAI since 2019 and served as its exclusive cloud provider.
Quickbase launched Pave, a full-stack AI application builder designed to address the "80% problem" in vibe coding, where AI-assisted development tools stall before reaching production-ready completion. The tool aims to integrate continuous quality validation into the development loop rather than ...
A Dev.to opinion piece describes a hypothetical "OpenClaw" scenario to illustrate concerns about vendor lock-in with AI code assistants, arguing that deep integration with proprietary AI tools could make migrating codebases technically or financially prohibitive.
An AI system processes persuasion signals — tone, conviction, repetition — as plain text, treating a senior developer's experience-backed objection with the same weight as a Stack Overflow answer. This means AI tools cannot serve as tiebreakers in deadlocked team decisions where emotional commitm...
Andrew Kelley, creator of the Zig programming language, stated that LLM-assisted code contributions are detectable because AI hallucinations differ fundamentally from human mistakes, and that Zig bans such contributions to its codebase.
Free daily email. Every story you just read, in your inbox at 6 AM ET.