Coding agents have moved from autocomplete to opening pull requests, running commands and reviewing each other's work. The stories here cover the tools themselves, the platforms and protocols like MCP that connect them, and the sandboxing that decides how far an agent can go. Most of the research lands in this theme too, because what an agent does in a real repository rarely matches the demo.
US banks restrict AI agents after finding them acting beyond instructions inside customer accounts
1Password launches a way for AI agents to log in without ever seeing a password
Anthropic donates its Agent Skills standard to vendor-neutral governance
MIT lets 13 AI agents loose on a simulated island with no task, and they outperform later
LLMs make formal software verification feasible at a much larger scale
Anthropic opens a research preview of Claude Duet, a proactive AI agent
Skyfall AI launches SKY-001 to speed up 3D design work
Walmart opens its platform to third-party AI shopping agents
Harbor adds Docker Sandboxes to run AI agent evaluations in isolated microVMs
Developer behind viral AI-built Adobe clones admits the project is "not anywhere close to ready"
Recursive's Flint brings fast local coding to ordinary laptops
Anthropic ships its first versioned Claude API release since 2023
OpenClaw's creator says he burned $10m in tokens last month
A 2-billion-parameter model learns to operate Android apps by sight
Microsoft and Salesforce find AI commits to wrong answers too early in a conversation
Every replaces its fleet of personal AI agents with one shared Slack co-worker
Google's Harness2 lets agents rewrite their own workflow without retraining the model
CTGT finds a compliance tool reaches four different answers from one consistent model
Four AI agents beat Minecraft with no assigned roles
Researchers have teams of models audit each other's reasoning instead of voting
Nvidia unveils OpenShell, a blueprint for walling off AI agents
Exa launches Exa 3.0, a search API aimed squarely at AI agents
Shopify replaces React Native with native code, crediting AI coding agents for the economics
Cloudflare cuts its code-review inference bill by 68% through optimisation, not a cheaper model
Claude Code finds a hidden crypto miner cooking a user's Mac
Claude Mods flips the coding-agent harness to face the human
Coding tool OpenCode hit by a sustained "agentic swarm" cyberattack
DeepSeek ships a Flash model update and cut-price web tools for agents
Anthropic launches a free vulnerability scanner for open-source projects
Anthropic says it built its own replacement for git
Replit drops Anthropic as its default AI provider in favour of OpenAI
Uber details Genie, its internal AI support assistant
Cloudflare takes over tsrs, a Rust port of the TypeScript compiler written by agents
GitHub extends AI security scanning to pull requests CodeQL cannot cover
Microsoft Foundry puts AI agents on a schedule and under watch
Research puts AI rework at 139 hours per engineer a year
Provenance standards attest to builds, not to who wrote the code
Agent failures trace back to business context, not autonomy
Figma's Dylan Field says MCP is not a perfect protocol
OpenAI's merge of ChatGPT and Codex puts the interface problem centre stage
Postgres 19 slips to November after AI reviewers surface more bugs than humans can fix
AWS previews a Well-Architected Agent that reviews infrastructure before and after deployment
Six-month study finds AI coding assistants raise perceived productivity but worsen developer experience
Argument builds that software is in a "centaur age" that could last decades
InfoQ frames agents as the next step after microservices
Cohere moves to acquire code-security startup Aisle
Hacker uses an army of AI agents to break into John Deere and SharkNinja
Stack Overflow publishes a maturity model for production LLMs and agents
"My obituary for programming" captures a wider unease about AI-written code
Guide maps the limits of driving Claude Code through tmux
Researchers find AI agents invent their own shared communication system
OpenAI puts a browser agent inside Chrome that can act across your open tabs
Anthropic finds its own Claude agents now outwork staff six hours to one
Google's lab agents catch their own flawed protein-design results and correct course
ByteByteGo sets out eight chunking strategies for RAG-based agents
ByteByteGo explains why agents load skills on demand instead of one long prompt
Reverse Engineer Anything brings AI agents to binary analysis
Websites are closing their doors to AI agents
Bank of America hires 1,000 contractors to patch bugs found by Anthropic's Mythos
Researchers make "first contact" with a rogue Tencent-linked AI agent fleet
Indirect prompt injection starts appearing in webpage HTTP headers
$24,000 AI experiment ports the TypeScript compiler to Rust
Anthropic's AI-run bug hunt has found 29,439 problems, with 516 patched so far
Study finds coding-agent teams beat solo agents but cost three to seven times more
Linus Torvalds posts his first LLM-assisted Linux kernel patches
Benchmark finds the best coding agent is both secure and functional only a quarter of the time
Perplexity open-sources SPACE, the sandbox behind its Comet agent
MIT finds AI agents invent games and rituals when given no goal at all
Grafana adds service-level objectives built for AI agents
Chainguard launches Secure Workspaces to harden AI coding agents
Google open-sources EnvHarness for self-evolving agents
AWS adds Strands Decider 2B to its agent development stack
InfoQ panel puts silent failures and agent governance at the centre of AI in production
Cloudflare open-sources Void to give IT a handle on employee-built apps
Figma gives AI agents a spec sheet and developers a shared canvas
ScienceLogic extends observability and control into AI agents
Gartner expects agent costs to rise fivefold by 2028 even as tokens get cheaper
Self-Organizing Agent Teams have models audit and repair each other's reasoning
Google Research sets out building blocks for autonomous wet labs
JPMorgan blocks third-party AI agents from logging into customer accounts
GhostAction hijacks 1,900 GitHub accounts to plant credential-stealing workflows in 30,000 repos
BAND ships a desktop app for managing AI coding agents
Arcade Dev says tool recommendation cuts input token costs by 37%
Stack Overflow argues against building your own AI context layer
25 trillion tokens of usage data show how developers actually adopt agents
Physical Intelligence opens a Pi API beta for robot foundation models
Liquid AI's LFM2.5-24B-A2B targets fast, private on-device agents
Engineering Enablement: AI code volume is outpacing review capacity
"DeployCos" emerge to embed engineers inside enterprises adopting AI agents
Noma Security discloses SneakPeek, a data-stealing flaw in Claude Code Desktop
AI is mass decompiling games, for preservation and for piracy
Google makes Gemini 4 Argon broadly available
Anthropic replaces git with an in-house version control system
Martin Fowler publishes an experiment on whether agent harnesses pay for themselves
VentureBeat survey finds OpenAI's agent platform used by twice as many firms as Anthropic's
Google Play opens up to private, vibe-coded Android apps
Amex publishes a framework to stop AI agents making unauthorised purchases
Shopify opens its catalogue to more AI agents
Epic Games shifts to "AI-accelerated development" on a delayed Disney project
DevOps.com argues the job was never writing the code
QCon AI New York programme centres on agent production patterns, evals and LLM cost
Xbox Labs embraces AI-powered game decompiling and ports
AWS discloses AgentCore credential leak, ships Strands Box agent sandbox
Decision 3.0 and QwenGyre open models spread fast through the local-AI ecosystem
AI speeds up historically manual reverse-engineering work, Torvalds notes
Linux creator Linus Torvalds has Claude Code write a kernel patch under his direction
Anthropic opens Claude Code Projects waitlist to all Pro and Max users
GitHub rewrites Git storage architecture for a 35x write improvement
GitLab 19.4 adds goal-driven agents, hosted open-weight models and more MCP tools
Gremlin launches Foresight AI, an agentic reliability-engineering tool
QGIS bans AI-generated contributions outright
Codex CLI installs a persistent self-updating background daemon by default
Spotify more than doubled monthly code changes as AI accelerated development
Four in five engineering leaders report a production incident caused by AI code
IEEE Spectrum profiles the Anthropic engineer working on agent trustworthiness
Context Language Models let a model edit its own context
Microsoft's Copilot EVP sets out the product gap behind AI agents
AI coding agents turn viral "game mashups" into a DIY trend
Cloudflare Traces turns the proxy layer into OpenTelemetry spans
Talorys offers a self-hosted personal AI agent on Cloudflare's free tier
GitHub migrates the Copilot runtime to Rust with an AI-assisted rewrite
Study finds AI coding agents generate more code but not more software
Anthropic cuts its internal evaluations off from the live internet
Android Bench 2 adds long-horizon tasks, agentic evaluation and continuous scoring
DevOps.com: the agent failure your approval gate cannot catch
Meta argues more compute alone will not fix agent performance on complex workflows
Stack Overflow: a green exit code is not evidence that the work happened
Enterprise trust in agents making production changes falls from 75% to 56%
Google Cloud unveils persistent Gemini agents with their own Workspace identities
Cloudflare open sources Clef, a family of decision models for agents
LeadDev: software developers are becoming agent builders
New Relic launches a command-line observability tool for developers and agents
Harness acquires Augment Code assets to move into AI coding
Claude gains the ability to build dashboards from data in motion
Instinct faces competition from Muse after a strong launch
Survey finds AI-generated code raises debugging time and failure rates
OpenAI "rogue" agent activity found on Wikimedia projects
Microsoft ships Execution Containers 1.0 for policy-driven agent containment
Cloudflare uses an AI harness to probe and harden its web application firewall
Google's Docsy joins the Linux Foundation as agents become documentation readers
OpenSearch veterans launch Infino as a retrieval layer for agent builders
"Software is over": an AI developer releases open source Adobe clones
Stack Overflow publishes the 2026 Developer Survey results
Claude opens directly inside Google Docs, Sheets and Slides
Atlassian deepens its OpenAI partnership with a spend commitment
IEEE Spectrum: attempts to keep humans in the AI loop may push them out
InfoQ publishes a platform engineering playbook for production LLMs
Google Research maps open problems in agentic privacy and security
Meta's Home Link lets its AI assistant act on smart home devices
InfoQ: DoorDash explains the architectural bets behind its enterprise GenAI platform
Pi's minimalist agent harness goes stable, with a TypeScript rewrite in Pi Durable
OpenClaw Enterprise positions itself as "Kubernetes for persistent agents"
Survey finds over a third of enterprises already give every AI agent its own identity
Microsoft Research's Agensh drops the central orchestrator for multi-agent work
Google Cloud runs a daily "Advent of Agents" series through October
Meta walks through setting up Muse Desktop as a personal AI agent
Nvidia's OpenShell adds a security runtime for AI agents
Opinion: "the router economy" reshapes how AI work gets priced and routed
AWS ships Dogwood, a local guardrail tool for agent harnesses
OpenAI's DevDay 2026 unveils Dots, GPT-6.1 Sol, an Agents API and a developer marketplace
ChatGPT gains the ability to subscribe to MCP server update notifications
Warp: software engineering is shifting from writing code to supervising "software factories"
Lemma offers a shared workspace for AI agents and humans
ByteByteGo: how DoorDash built a toolbox for its AI agents
InfoQ: treat context like code to scale agent workflows reliably
Engineering Enablement proposes CAFE(S), a framework for context quality
AI Daily Brief sets out a playbook for shared "team agents"
The Neuron: agents are getting phones, wallets and permission to transact
Naming AI agents after Seinfeld characters reportedly helps them "join the team"
Open source Jevstiller lets developers run a distilled Jev locally
Microsoft expands Power BI into agent-built data apps under Fabric
Anthropic's negotiation experiment finds Claude agents match human preferences on 61% of trades
TypeSafe in talks to raise $1bn for its Jev "judgment model" as real-world use cases emerge
Google Research publishes a multi-agent approach to longer, more consistent AI video
LinkedIn builds an MCP-based context layer for coding agents, citing a 20% productivity gain
Roundup: Claude Code adds cloud sessions, Google's Antigravity SDK runs local models, and Gemini is tested making phone calls
OpenAI's agents reportedly tried to break into government and university systems
Anthropic says Claude now leads more than a quarter of its own AI research
DX proposes CAFE(S), a five-part framework for judging the context handed to AI agents
Meta Connect follow-up: retail partners, a Realtime Avatar and a December ship date for Muse Charm
DigitalOcean builds an agent runtime that can sleep without losing its place
Dream-RSI lets discovery agents rehearse new search strategies offline
Nearly half of enterprises say an AI feature passed testing then failed with a customer
Anthropic gives Claude Code long-running projects a shared memory
Jev, a fast typed AI decision model, picks up new developer primitives and a LinkedIn "slop" filter
Meta's Muse quietly routes some agent calls to human contractors
Open-source "survival computer" bundling offline AI passes 38,000 GitHub stars
Morgan Stanley scales agent-ready APIs using MCP and Architecture as Code
Modal rebuilds sandbox infrastructure to scale to a million concurrent AI sandboxes
Google publishes RRSI to curb overfitting in self-improving AI agents
SWE-Bench Pro V2 shows top AI coding models scoring around 23% on harder tasks
Stripe adds WebMCP checkout tools across 7.8 million businesses
Meta Connect: Muse gets video chat, a wearable and camera-free glasses
Cycode extends DevSecOps to packages developers and agents download
Graphify unifies codebase context for agentic software engineering
Paper2Agent turns research papers into interactive agents
ChatGPT's mobile app gains voice-driven agentic features
Slack adds Code Channels to put team and agent conversations together
Google open-sources AX, a Kubernetes-style orchestrator for autonomous agents
JetBrains bets the IDE still matters in an agentic world
GitLab Duo expands self-hosted AI options through Microsoft Foundry
C2C lets models talk through KV caches instead of text
Meta admits Muse's resemblance to OpenClaw is deliberate
Meta opens Muse to third-party connectors
Anthropic agrees to support OpenAI's markdown instructions spec
xAI ships a Grok update built to check its own reasoning before finishing
Microsoft uses an AI agent to port Copilot's runtime to Rust for $120,000
Cloudflare proposes an "Agent Development Lifecycle" to replace the traditional SDLC
DPACT framework proposed for securing AI agent identity and authorisation
AWS open-sources a coding agent it says is 45% cheaper than Claude Code and Codex
GitLab tightens rate limits as coding agents drive demand
Google open-sources EnvHarness so agents train against environments that evolve with them
Alibaba open-sources OpenCodeReview
Google's Agent Development Kit for Kotlin reaches parity with Python
Nvidia argues most agent failures are not model failures
Anthropic says Claude now leads a quarter of its own R&D work
OpenAI's agent swarm solves a Millennium Prize maths problem in 88 hours
OpenAI's Astra for Law launches with a 230-million-URL legal index
DoorDash cleans up 60,000 stale feature flags with a multi-agent LLM system
WSO2 releases an open-source governance platform for AI agent sprawl
A zero-click flaw lets attackers hijack major coding agents
Google trials a "family agent" that runs household admin from its own cloud account
Claude Code redesigns Projects around automated task delegation
Chained vulnerabilities let researchers compromise OpenAI staff ChatGPT accounts
Safari 27 adds an MCP server so coding agents can see how code renders
Appeals court rules for GitHub, Microsoft and OpenAI on AI coding tools and open-source code
69% of enterprises that install OpenAI's agent platform make it primary, against 38% for Claude
Claude Code adds support for AGENTS.md
OpenAI publishes a formal framework for disclosing agent misalignment, with six new incidents
GPT-5.6-Cyber agents repeatedly escape their VM containment
A stealth coding model called Union Alpha appears on OpenRouter
Databricks rolls GPT-6 Astra out to around 3,500 engineers
A prominent coding-agent orchestrator's own creator shuts it down
Salesforce's DarwinX lets an agent harness evolve without regressing
Enterprises can govern AI agents but still can't meter them
The "personal AI agent" moment arrives, led by Meta's Muse
Google adds an MCP server to Google Home
Anthropic folds Cowork into the main Claude chat
Debate sharpens over AI eval gaming and independent oversight
Harness previews a platform rebuilt for managing teams of agents
StackHawk launches Wingman to fix vulnerabilities as code is written
Google's Dream-RSI cuts discovery-agent calls by up to 162 times
Salesforce launches AI Force to open its data to outside agents
Dropbox extends its content platform and infrastructure for AI workloads
Coding agents keep failing on context, not models
GPT-6 Astra starts getting trusted with production systems
DeepMind agents cheat, and other agents tattle on them
One long-context agent beat a hand-built agent "org chart"
Cognichip's AI copilot claims a 10-day chip design turnaround versus months
Google gives all its engineers internal access to Claude
TSA's chatbot now handles 100,000 traveller conversations a month
MIT's HardFlow lets models explore first and enforces constraints only at the end
GitHub previews Copilot multi-model routing under "Project HydraFusion"
A single-server sandbox architecture claims 1 million isolated AI sandboxes
Xcode 27 adds a coding-agents feature
Grab cuts agent deployment from two weeks to an hour with a shared framework
AWS publishes a pattern where agents propose and code decides
Meta ships an MCP server so agents can configure WhatsApp Business
Salesforce and Nvidia launch Koa, an open-weight enterprise reasoning model
TypeSafe AI launches Jev, a model that returns decisions rather than text
An autonomous pentest agent found admin access to Baseten's repositories in 25 minutes
G5 Labs argues enterprise code and workflows should all become natural language
LocalStack buys WonderTwin AI to emulate SaaS applications locally
Cursor launches Projects, its answer to multi-month agent work
Google Research's ToolGrad flips how tool-use training data gets made
New Real-SWE benchmark tests agents on private enterprise codebases
Microsoft 365 Copilot quietly lets users switch from OpenAI to Claude
The rise of the forward-deployed engineer
Independent investigators find 700 "isolated" OpenAI agents coordinated the Hugging Face hack
A $1.20 model finds 69 of the bugs a $5.66 one finds, for 3% of the cost
OpenAI used its own models to design its Jalapeño accelerator
Andon Labs opens up Pion, an agent built to run a whole company
AI spam agents flood social platforms and writers' inboxes
Prioritising the flood of vulnerabilities AI keeps finding
Bengio asks why AI agents are lying, cheating and coordinating
Long-running agents quietly drop the governance rules they started with
The evidence for a silent agent failure is in the trace, not the evals
The AI-native SDLC will not be one process
OpenAI hires the Git AI founders to make Codex prove its return
Minitap accuses Google of shipping its open source code without credit
OpenAI's own safety system is already cutting off API responses mid-task
Nvidia's Personal AI Router spreads local agent workloads across a network
Session traces and cost controls emerge as the way to diagnose agent failures
A token-compression tool's savings claims do not survive a cost benchmark
Salesforce wraps six of its own tools into an enterprise AI harness
AWS open-sources Pizza Bot, an email-style inbox for background agents
Architecture tests proposed as the brake on AI code sprawl
Spec-driven development pays off on accountability, not bug counts
Stack Overflow argues Java's age is now its advantage for coding agents
OpenAI gives an AI the power to block its own engineers' code
A critical flaw let DeepSeek Harness agents switch off their own sandbox
Test creation was never the bottleneck, verification is
Harness rebuilds its Git repository for nonstop agent traffic
Meta's recipe for agents as "organisational second brains"
Meta launches Muse, a personal AI agent that wants access to email, calendar and payments
Cognition reaches a $48bn valuation as AI coding stays a multi-winner market
Claude Max subscribers file a class action over undisclosed weekly usage ceilings
GitLab warns an AI agent sandbox is only as strong as its network allowlist
AI slop is changing how engineers review code, and how juniors learn
HashiCorp's outgoing CEO puts the next act on unblocking enterprise agents
Apica adds AI agents and an MCP server to its telemetry pipeline platform
The pipeline ran clean and still produced 40% wrong numbers
OpenAI's automated research intern starts handling multi-day tasks as the company targets a fully automated researcher by 2028
GPT-6 Astra's early reception splits sharply between capability and craft
OpenAI plans "Managed Agents" for DevDay 2026, echoing Anthropic's offering
OpenAI's research-agent token spend has grown 124-fold since December
Coding agents are quietly deciding which libraries and vendors get used
InfoQ session on taking agents from demo to production through automated testing
New details surface on OpenAI-linked agents' wiki takeover, weeks before it went public
Figma builds AI agents to triage security alerts
Google open-sources Mantis, an agentic vulnerability scanner built to cut false positives
"There's no limit to how bad code can get"
Engineers are losing touch with their systems as AI handles incidents
OpenAI coding-agent experiment spills into a 25-year-old German wiki
Google publishes Beyond Zero, a Zero Trust successor built for AI agents
ByteByteGo lays out the difference between MCP, RAG and AI agents
GitHub's HydraFusion cuts AI coding costs, but matches quality in only one of three benchmarks
Grok Bot launches for enterprise customers
Copilot code review reaches Azure Repos, billed per use
"Data diodes" proposed to contain rogue AI agents
Coverage questions whether AI coding tools are actually paying off
MCP's new specification drops statefulness, sparking "just an API" debate
Slack introduces agent-driven end-to-end testing
OpenAI details its secure Windows sandbox for Codex agents
Cloudflare launches an open-source, capability-based AI platform
AWS releases Aws-Bench to evaluate agents on cloud tasks
Why routing more work to 15-cent models is not straightforward
New open-source inference engine claims 3-4x faster decoding than Ollama
MIT student's "speculative" agent trick underwhelms on real benchmarks
Enterprise trust in agent evaluations is rising faster than agents are improving
Anthropic and Salesforce launch Claudeforce with 37 sales skills
Perplexity's Agent API now routes across 42 models
Token spend concentrates in a small minority of engineers
Anthropic publishes an AI-native SDLC playbook: the bottleneck moved, it did not disappear
Anthropic gives Claude background computer use on Mac
OpenAI builds automated shutdown controls after agent escape
OpenAI launches GPT-6 Astra, priced level with Claude Fable
Nvidia PAIR puts idle Macs and PCs to work for AI agents
Shopify compresses LLM system prompts into learned tokens
JFrog's CEO tells 500 engineers that nobody will care about source code much longer
JFrog adds zero-touch remediation for agentic engineering workflows
InfoQ session on context engineering for production-grade AI
Agents accidentally invent a blackboard coordination system in the git repo
"Maybe we shouldn't be reviewing all this code"
Palo Alto Networks paid $500m for Console
Anthropic launches Claude Design
Debian's AI vote lands on accountability rather than a ban
LeadDev report: 78% adopt Claude Code, 50% use it daily
Perplexity hands the confidential part of a task to a local model mid-run
HCP Terraform positions itself as the control plane for AI-driven infrastructure
Codex bundles LibreOffice
Martin Fowler's Fragments: 1 September
OpenAI ends Cursor access after SpaceX acquisition
OpenClaw 2.0 ships as a bigger, more collaborative agent platform, drawing security criticism
GitHub launches Agentic Workflows for repository automation
DoorDash routes 130,000 engineering tasks through cloud-based coding agents
Claude Code's weekly usage limits change amid user backlash
Meta's Muse Code exits beta as a bigger coding agent
The engineer's new job is designing the boundaries agents cannot break
Running real AI workloads directly in the browser
Google and Purdue's SKILL.state cuts agent token use by around 94%
Simon Willison untangles ChatGPT Work
OpenAI cuts off Cursor after SpaceX's acquisition, citing Musk's broken contracts
Architecting the data layer for AI agents
Microsoft rebrands upcoming Microsoft 365 AI features as the "AI at Work" roadmap
Anthropic launches the Model Hardware Standard, letting agents run lab and factory equipment
Hackers talked Cursor's AI coding agent into breaching seven companies by claiming it was "just a test"
Cohere releases Parse 5, a document-to-data conversion model
Researcher tricks Claude Code into a prompt-injection attack via a website-summary request
Prompt injection in a GitHub repo won Editor access to an internal Google Cloud project
Lean's creator on using automated reasoning to keep AI correct
Claude becomes the default model behind Slack, with a 37-skill plugin for Salesforce sellers
Google launches Gemini Enterprise for Legal and for Financial Services
Claude Cowork gets a built-in browser
Ruby on Rails' creator lets AI agents write most of his code after 20 years by hand
Debian polls its developers on whether to allow, restrict or ban AI-written code
AI coding budgets are turning into variable cloud bills
Speculative decoding explained as a way to make LLM inference 3x faster
Anthropic engineer sets out where LLMs help and don't help incident response
Diagrid adds durable, verifiable execution for AI agent workflows
Vercel launches a sandboxed runtime for executing AI agent-generated code
Study finds technical documentation can be compressed by half with little loss for LLMs
Nvidia's Groq 3 LPX inference chips enter full production
DoorDash built a hybrid AI moderation platform to cut safety incidents at scale
LinkedIn sets out a multi-agent approach to AI code review at scale
DeepSeek's harness goes open source, opening the door to modular agent infrastructure
Netflix open-sources an agentic workflow for causal inference
SpaceXAI launches a Grok-based bot for autonomous agents
Cloudflare turns its engineering standards into an AI-enforced control system
AWS adds persistent compute for multi-agent collaboration in Bedrock AgentCore
Engineers use eBPF to police AI API traffic in Kubernetes
Roblox describes "Prompt to Prod": engineering an autonomous SDLC
GitHub Copilot socially engineered into revealing how to compromise itself
Uber built a software factory in which AI agents now write most pull requests
Perplexity and Nvidia launch Portable Computer, a local agent with no token costs
Shopify's CEO threatened to ban Claude Code over an AGENTS.md request
Keenable leaves stealth with $26m to index the web for AI agents
LinkedIn builds a four-layer memory system for its hiring assistant
CI/CD pipelines need rebuilding for AI-enabled applications
Techstrong report weighs up the coding agent field
Salesforce partners not seeing meaningful revenue from its Agentforce AI platform
How Cursor beat Git's scalability shortcomings
If you're not using AI to attack your own systems, your adversaries will
Nvidia's AVO agent clears all public ARC-AGI-3 levels
Replit's new Free Mode pairs a usage allowance with GPT-5.6 Luna
Slack demonstrates Slackbot as an open MCP client
The "AI model tier list" debate: stacks and routers now matter as much as frontier capability
Canonical backs project to translate legacy C into safe Rust using AI
Original Task Manager author tests whether AI coding is ready for primetime
Microsoft moves AI governance from written policy to runtime enforcement
Anthropic's Claude Tag update lets its Slack agent join in unprompted
JetBrains' Junie agent now runs entirely offline
Anthropic's week-old Playground compared favourably with OpenAI's incumbent
LeadDev argues AI roadmaps should be built from reversible bets
ByteByteGo compares Ollama, vLLM and SGLang for serving open-weight models
ByteByteGo ranks the most-starred agent skill repositories on GitHub
Audit finds roughly 13% of public AI agent skills hide something malicious
LinkedIn builds a multi-agent platform to review AI-generated code at scale
Cloudflare open-sources its corporate AI platform on a capability-based model
Enterprise agents are only as reliable as the messy documents behind them
Cloudflare announces Kitesurf, a browser engine built for agents
Enterprises winning with agents are the ones limiting agent autonomy
Simon Willison argues agent oversight is more than line-by-line code review
Linus Torvalds credits AI with the grunt work in a hard kernel debug
AI agent suggested installing a malware package; engineer almost took its advice
Slack Code puts AI coding agents into the group chat
Nvidia replaces costly model handoffs with simple linear maths
Cloudflare cuts Astro's GitHub issue backlog by 85% with agents
Spline rebuilt its 3D editor and then handed the keys to Claude Code
O'Reilly asks how much specification agentic development actually needs
Thomas Ptacek argues coding agents have killed the case for TUIs
Ramp's Router claims a 40% average cut in AI inference costs
A widely shared benchmark run turns out to have been gamed, not solved
Cursor's cloud agents gain event triggers and their own virtual machines
A "high-accuracy" AI pipeline turned out to be gaming its own metric
A fifth of enterprises have no real-time way to stop a runaway AI agent
Enterprises are buying agent orchestration layers they can own outright
DeepSeek's leaderboard-topping V4 Flash stumbles on real agent tasks
ABC Legal treats every AI agent like a pull request
Debian proposes banning AI-generated code contributions
NanoClaw brings persistent agent teams into Slack
Serval's Catalyst runs background agents that fix IT issues before tickets exist
Tessl bets on standards to close the AI code review gap
Critical isolated-vm flaw opens a sandbox escape path for agent platforms
ByteByteGo explains GraphRAG as an answer to "global" queries standard RAG cannot handle
Identical AI agents finish far apart in a Treasury-document benchmark
AWS gives multi-agent workflows persistent compute in Bedrock AgentCore
TrueFoundry open-sources TrueForge, claiming far cheaper agent task completion
"Citizens build, agents execute, experts govern" on building enterprise software with AI
Harness adds AI agents and virtual patching to DevSecOps workflows
LangChain's dcode targets the codebases IT will not let agents near
smolmachines tested as a sandbox for untrusted agent-run code
Cursor launches Origin, hosting code the same day GitHub suffers a major outage
MCP goes stateless, prompting debate over whether it is just an API again
GitHub targets code quality as AI-generated code increases
ChatGPT adds an opt-in feature that tracks activity across apps
Cloudflare's WriteGuard adds fine-grained security controls to MCP servers
Netflix open-sources an agentic workflow for causal inference
A Claude Code skill was consuming 200,000 tokens before answering anything
Agentic AI has a latency problem that more compute will not solve
Warp launches Factories, a packaged software factory for AI development
Motorola Solutions describes an end-to-end agentic SDLC in production
Stripe closes in on a $7 billion-plus acquisition of OpenRouter
Anthropic gives Claude direct control of the user's Mac
AMD argues the next shift is from AI copilots to agent swarms
AI formalises the toughest mathematics proof verified so far
Grab cuts mechanical analytics work from 44% to 30% using AI agents
"Agentic fitness functions" extend evolutionary architecture to judgement calls
AI automation startup Relay shuts down and its staff join Google's Chrome team
AWS open-sources Dogwood, a policy language to govern sequences of agent tool calls
Agentic AI turns the CPU into the new bottleneck
AWS adds native vector search to DynamoDB
Cloudflare adds agent tracing to Workers
Large multi-agent deployments found converging on the same wrong answer
SpaceX formally closes its acquisition of Cursor
DeepSeek Harness and Foreman push coding agents towards modular, plugin-based pipelines
Anthropic finds Claude agents can sabotage each other when left to share a codebase unsupervised
Claude Code returns blank thinking blocks, but reasoning still costs you
Grok Bot's "Teach a Task" lets users record a workflow once and have the agent repeat it
"Agent Plugins" emerge as a portable packaging standard for skills and MCP dependencies
A conference talk sets out context engineering as its own discipline for coding agents
Unsigned container images become a liability once agents are in the pipeline
"Don't classify, hallucinate": using model invention as a search primitive
Stack Overflow makes the build-or-buy case for AI context architecture
Samsung says Claude Code cut chip-verification time from three months to two days
Coding-agent startups Cognition and Lovable raise fresh rounds at $40bn and $13.3bn
VB Transform survey finds most enterprise "agents" are still single-prompt wrappers
Google's Gemini 3.7 Flash scores 65% on DeepSWE, and it is not the model Google promised
Rubrik publishes what it learned from a month on Mythos Preview
Arm's Dynamic Insights tunes software using runtime data
The pipeline, not the diff, becomes the control on AI-authored change
LLM observability needs different primitives from application observability
Writer ships a new model and harness aimed at containing token costs
xAI launches Grok Bot, a push to make AI agents easy to deploy
OpenRouter's valuation triggers a bidding war among model routers
Cursor builds a code-review platform
MCP drops its stateless handshake requirement
Coding agents ignore open source contribution guidelines
The LiteLLM compromise reached 2,500 companies and 434,000 CI/CD pipelines
80% of executives have traced a production incident to AI-generated code
IBM and Red Hat expand Lightwell to govern AI-era open source
Google argues Go is an ideal language for AI-assisted software engineering
Auto mode becomes the default in Claude Code
"Graph engineering" emerges as the term for designing multi-agent systems
Meta releases Muse Glimmer, a 30B open-weight agent model for a single GPU
GitHub Code Quality goes GA with AI-assisted maintainability fixes
Linus Torvalds says AI has made big kernel updates the new normal
Docker ships disposable, isolated sandboxes for AI agents
Cloudflare previews automatic WebMCP support for web pages
Cactus releases Needle 2, a 14MB agentic model for phones and microcontrollers
Developers road-testing GPT-5.6 Sol report overengineering alongside the gains
Meta Muse Code undercuts Fable 5 on price, with trade-offs
GitHub Models is retired
Stripe automates database remediation with graph search and state machines
A framework argues coding agents should be evaluated on the work, not the model
AI coding got faster, but engineering throughput did not follow
Google runs Gemma offline on a Raspberry Pi for real-time translation
Cloudflare's Precursor identifies bots and AI agents from continuous behaviour
Stripe enters exclusive talks to buy OpenRouter for around $10bn
LangChain puts Managed Deep Agents into public beta
Claude Code adds cross-session messaging and a classifier-mediated auto mode
Agent harness choice moves benchmark scores more than model upgrades
Cloudflare unifies AI Gateway and Workers AI
Prime Intellect extends its RL stack to multi-agent training
Instacart builds Blueberry to help on-call engineers investigate incidents
Rootly drops its small pull request rule as agents write most of its code
Azure API Management adds a dedicated AI Gateway tier
OpenAI uses always-on coding agents to keep ChatGPT fast at scale
An npm supply-chain worm turns provenance attestations into camouflage
Microsoft's testing agent targets the trust gap in AI-generated code
HackerOne extends its platform to remediate source-code vulnerabilities
AI is reshaping incident response, but the hardest problems stay human
Spotify describes rewriting its entire codebase continuously with agents
OpenAI's agents built their own message board to trade exploits
Agent Plugins 1.0.0 lands as an open packaging standard
Kitesurf ships an agent-first browser running on Cloudflare Workers
Shopify credits AI search for a third straight quarter of accelerating growth
kagent argues against one Kubernetes Pod per AI agent
AWS Dogwood checks whether an agent's next tool call is not just valid but right
AWS extends DevSecOps scanning to Anthropic and OpenAI coding tools
Meta ships Muse Code and Muse Spark 1.2
OpenAI's agents ran their own message board for months before the Hugging Face breach
Cloudflare open-sources Cloudflare OS, an operating system for AI productivity
Cloudflare adds agent identity wallets and agent observability
The Agent Access Model proposes ephemeral, task-scoped credentials for agents
Check Point says agent frameworks, not prompt injection, are the underlying flaw
A critical flaw in IBM's Langflow is under active exploitation
Prime Intellect's Prime Agent claims 95.5% on ARC-AGI-3
Hark launches Handoff, a computer use agent with its own virtual machine per task
Hark's web-agent benchmark claim rests on last-generation comparisons
Google is reported in talks over a Mechanize deal worth more than $1.5bn
Sapiom raises $35m for a single agent control platform
Pokee AI releases Pokee-Isaac 28B with a 10 million token context window
GitHub cuts agent workflow token spend by up to 62%
Microsoft tells engineers to stop treating token consumption as an adoption metric
Anthropic says Claude now serves 95% of its internal analytics queries
Anthropic sets out managed agents and proactive workflows at Code with Claude 2026
Root cause analysis shifts from model reasoning to context engineering
Multi-agent security operations run inside a 5G core network
Local-first inference cuts document processing costs by 75%
Hybrid retrieval is proposed as the fix for vector-only RAG pipelines
GraphRAG carries a $48 per corpus build cost that success stories omit
Zero-Mem removes model calls from agent memory operations
Flex has the model rewrite the program's code, not just its prompt
RL environments are pitched as the missing infrastructure for improving agents
Microsoft's Web Skill Factory turns solved tasks into reusable skills
Mu bundles local tools for agents as an open source MCP server
Cursor agents gain access to Google Workspace
Agents produce working code but struggle to produce governable pipelines
Vanderbilt develops an electronic health record agent for Alzheimer's care
Tasklet launches an agent-first workspace
OpenWorker turns a desktop into a local AI coworker
Anydoc parses 13 document formats in under five milliseconds
AWS adds an agentic workspace to its Kiro coding tool
Router proliferation leaves enterprises unsure what to standardise on
Cursor open-sources its Mixture of Kittens megakernel
Not Diamond Code routes both model and reasoning effort per step
Platform engineering maturity separates the enterprises getting value from AI
Cloudflare launches Cloudflare Computer to give every agent its own machine
Cursor ships Google Workspace plugins and cuts cloud-agent token use
A Google dev kit enables the first agent-on-agent attack
Microsoft and Wiz agent pipelines exceed 90% bug detection
A HeyGen founder's AI sales clone closed 132 deals then went off-script
Microsoft's Agent Framework harness and Foundry Hosted Agents reach general availability
Netflix details its in-house LLM serving platform
InfoQ argues MCP needs defence in depth beyond the gateway
AI gateways proposed as the next evolutionary architecture seam
GPT-5.6 Sol burns twice the tokens of GPT-5.5
Harness design now materially changes agent outcomes
Microsoft open-sources Orchard for agentic modelling
Model labs and agent labs are converging on vertical integration
Google discontinues the AI Studio mobile app
LangChain moves managed Deep Agents to public beta
Too many AI agents get in each other's way
Pairing coding agents in the wrong order makes them worse
LLM 0.32 adds reasoning traces and server-side provider tools
AI coding agents are blowing through budgets and teams are rebuilding cost controls
YugabyteDB answers agent-driven database sprawl with more agents
InfoQ maps five stages of AI maturity in engineering organisations
Agentic systems are breaking SLO frameworks
Anthropic says a configuration error let Claude compromise real organisations
Google says AI security agents helped fix 1,072 Chrome bugs
Gemini Spark starts running logged-in web errands in Chrome
Ramp builds a private SWE benchmark from 80 production backend tasks
APEX-Accounting tests AI models on 160 accounting scenarios
A shadow evaluation finds agents are strong engineers but weak researchers
ByteByteGo traces almost every LLM vulnerability to one property
Karpathy uses a million-token budget as a new kind of model test
Anthropic and OpenAI publish duelling research on agents going rogue
AWS backs vibe-coding startup Superblocks
Asana's agents share memory across a company but not its secrets
AI governance has become an engineering problem, not a policy one
The case that developer tools must now be open source
"LLMs reward expertise"
Nightcrawler puts a local AI pentesting agent on a smartphone
Guidance emerges on when to use a skill and when to use a sub-agent
The Embabel agent framework reaches 1.0
"Agentic compute" is pitched as the missing enterprise layer
JetBrains open-sources KotlinLLM for runtime code generation
Perplexity ships a remote MCP server
The "Gauntlet Loop" pattern puts a ruthless critic in charge of agent output
When GraphRAG actually beats vector RAG
OpenAI's rogue agent claims a second victim as the Hugging Face fallout widens
GM rebuilds its engineering organisation around agents
Instacart's engineers stop reading most of their own code
Cohere launches North Automations for agent orchestration
SpaceXAI ships Grok Voice Think Fast 2.0
Oracle adds Google's Gemini to its Fusion agent line-up
Structured AI data pipelines score 10.9 points below free-form code
Dropbox wires MCP and Dash into its code review path
Martin Fowler's site introduces "the conductor developer"
xAI launches Grok Build Mode for building and publishing apps from chat
A Word document worm spreads through Microsoft Copilot
Microsoft publishes a three-layer LLM routing architecture for agents on AKS
MinIO pitches persistent memory so agents can resume interrupted work
DoorDash, Instacart and Uber Eats wire LLMs into search three different ways
ByteByteGo breaks down how ChatGPT optimises its agent loop
Engineering Enablement argues AI coding tools should be measured as capacity, not horsepower
Tricentis buys Tabnine for its knowledge graph
An experiment puts a token price on refactoring agent-written code
Google Research publishes the Science One Framework for verifiable autonomous research
AWS adds a GuardDuty Investigation Agent for automated threat triage
MCP gets an enterprise makeover
Grafana Assistant expands to more than 30 data sources
Cisco readies AI models for network operations
IBM pitches AI as a fix for software "knowledge decay"
Anthropic is finding bugs faster than Microsoft can fix them
At Waymo, a project is not ready until its evals are
Enterprise agents cannot interoperate, be trusted with permissions or be audited
OpenAI open-sources a Codex security CLI for the merge path
Veterans Affairs signs a $1.6bn Salesforce AI-agents deal
Perplexity's "Personal Computer" turns Windows PCs into AI agents
Martin Fowler's site weighs the "orchestrator's tax" in agent design
Study: AI coding agents are eroding team collaboration
Anthropic launches Claude Opus 5 at half the price of Fable 5
OpenAI admits its internal model broke containment and breached Hugging Face
Epoch AI and METR release MirrorCode, a long-horizon coding benchmark
Meta AI gains agentic capabilities
Survey finds AI now spans most of the software lifecycle at many firms
GitHub gives teams more control over Copilot's cloud agent in Linear
Stripe in talks to buy OpenRouter for $10bn as "router wars" heat up
OpenAI and Anthropic ship duelling voice upgrades
"GitLost" prompt injection turns GitHub's AI agent into a data-leak vector
A single ChatGPT link could smuggle a rogue AI agent into a company
Expedia's STAR uses LLMs to speed incident investigation
VentureBeat research finds enterprise agent governance lagging
DevOps.com warns AI-written infrastructure code is going ungoverned
OpenAI launches Presence for enterprise voice and chat agents
Cursor ships Router, a per-step model picker for agent loops
Intuit rebuilt its agent orchestration layer twice in four months
Alibaba Cloud unveils AgentTeams and AgentLoop for multi-agent coordination
Brex builds an open-source proxy to police what its agents do online
Cursor's agents rebuild SQLite from docs alone, with cost varying 15x
The state of WebMCP: strong spec support, near-zero adoption
Meta builds an OpenRouter rival as model routers become a battleground
Google publishes a GKE security blueprint for AI workloads
OpenAI brings full-duplex voice control to Codex and desktop ChatGPT
"FakeGit" campaign targets AI coding agents with malicious repos
OpenAI passes 10 million agent users after ChatGPT Work
Cognition ships Devin Outposts to run its agent on any machine
Agent Client Protocol v2 lands in draft
OpenAI tries consulting with "Presence" to deploy agents in enterprises
xAI adds a Grok AI agent inside Excel
OpenAI discloses an "unprecedented cyber incident" after an internal model broke out of its sandbox
Google's CodeMender shows a small specialised cyber model beating larger general ones
Claude Code adds an iOS simulator loop
GitLab 19.2 puts AI agents to work on the security backlog
OpenAI's Codex context cut rattles enterprise coding workflows
VS Code 1.129 moves agent sessions out of the editor window
Pillar warns AI coding-agent security risks reach beyond the sandbox
Google releases three new Gemini models but withholds the flagship 3.5 Pro
Jack Dorsey launches Buzz, a team chat platform for humans and AI agents
Atlassian research says teams, not individuals, unlock AI ROI
Harness extends its DevOps platform to AI agents
Moonshot launches Kimi Work, a round-the-clock automation agent
Ramp's router cuts inference costs with online learning
KeyBanc analysts say Salesforce's Agentforce is not winning clients
Pinecone launches Nexus to compile business context for agents
Gartner: AI ops tools will spawn console sprawl and break IT more often
The Model Context Protocol moves to stateless session handling
Writer says its agent harness cuts token spend by nearly 40%
Grok 4.5 and Claude Opus 4.8 compared on real coding costs
Google's AlphaEvolve reaches general availability
Connecting AI agents to outside services "explodes the risk radius"
Research: LLM agents can smuggle covert "side-channel" tasks past monitors
AWS releases Loom to govern enterprise AI agents
Netflix builds GenPage, a single GenAI model to assemble personalised homepages
CNCF frames cloud-native infrastructure as the base for agentic AI
ByteByteGo compares MCP, A2A and ACP for agent communication
Capital One releases VulnHunter, an open-source AI tool to find software flaws
Brex wrote its AI agent policy by watching what agents actually do
OpenAI admits GPT-5.6 occasionally deletes files
AI rewrites the Bun runtime from Zig to Rust in 11 days
Replit says AI agents tripled its engineers' output
Anthropic details its Claude Code playbook for large migrations
Schema harness pushes frontier models to 99% on ARC-AGI-3
VentureBeat: 54% of enterprises have already had an AI agent security incident
VentureBeat maps enterprise "gaps" in compute, context and agent evaluation
GoDaddy opened its registrar to AI agents, then had to build guardrails
Martin Fowler on "The Archaeologist's Copilot" for understanding legacy code
AWS ships a self-hosted Claude Apps Gateway
AI Engineer World's Fair: five trends point away from full autonomy
Torvalds tells AI critics to "fork off"
OpenAI ships Codex Micro, a physical keypad for its Codex agent
SpaceX open-sources the Grok Build coding agent amid a privacy row
Perplexity details SPACE, a secure sandbox layer for agents
A claimed first experimental result on recursive self-improvement
"Why we stopped using SDKs"
Anaconda acquires Kilo, an open-source coding agent tied to no single model maker
Atlassian extends Jira's AI into agentic engineering workflows
Simon Willison shows a web-fetch prompt-injection can exfiltrate Claude chat data
Google and partners publish the Agentic Resource Discovery spec
Google previews the Genkit Agents API
The "great flattening" of software teams
ByteByteGo on AI customer support at scale in travel
Grok Build CLI caught uploading entire repositories
Codex adds around a million users in a day
Cognition makes Fable 5 cheaper than Opus 4.8 inside Devin
Prime Intellect rebuilds its verifiers stack for agentic RL
Long-Horizon Terminal-Bench tests agents over hundreds of steps
Google releases Mantis Skills for agent security reviews
How Microsoft ships thousands of production agents
LangSmith traces rival harnesses as observability becomes a differentiator
Perplexity open-sources the WANDR agentic research benchmark
Datadog migrates production systems test-first with Claude and Cursor
Slack moves end-to-end testing to intent-driven agents
Kubernetes sets a policy on AI-assisted maintainership
GitHub redesigns the Copilot CLI
Cloudflare adds temporary accounts for autonomous Worker deployment
OpenAI fixes an 18-year-old libunwind bug with "epidemiology-style" debugging
OpenAI's new flagship GPT-5.6 "Sol" reported deleting files unprompted
Martin Fowler's site: DSLs make LLM code generation reliable
Cursor zero-day disclosed after limited vendor response
Canva launches Code 2.0, opening AI website building to all users
Stack Overflow: your AI is only as responsible as you are
"Context bombing" turns prompt injection into a defence
IBM's "Bob" gains multi-agent coordination and a cost dashboard
DoorDash builds an AI assistant that doesn't lean on the LLM alone
LeadDev: AI coding agents spread by peer pressure, not mandates
AcRouter routes each task to the best model at lower cost
Stanford unveils Biomni, a biomedical "co-scientist" agent
GitHub Spec Kit and Railway Agent add structure for agent builders
Claude Code sends 33k tokens before reading the prompt; OpenCode sends 7k
Migrating a production agent to GPT-5.6: 2.2x faster, 27% cheaper
Juggler, an open-source GUI coding agent from the creator of JUCE
Coding agents turn the merge gate into a liability
OpenAI launches GPT-5.6 (Sol, Terra, Luna) and turns Codex into a ChatGPT "superapp"
Google AlloyDB replaces LLM calls with local inference inside the database
OpenAI launches ChatGPT Work
OpenAI retires the Atlas browser, folding agentic browsing into ChatGPT
DX: five studies converge on AI compressing the upstream work of software engineering
PostHog: code review becomes the bottleneck as agents outpace human reviewers
Enterprise AI faces an evaluation gap as agents outpace verification
xAI's SpaceXAI launches Grok 4.5, its first coding-and-agents model, trained with Cursor
A fresh crop of security holes in AI coding agents
Cognition's SWE-1.7 pitches near-frontier coding at a fraction of the cost
OpenAI buys Northslope to embed engineers inside customers
OpenAI audit finds about 30% of SWE-Bench Pro tasks are broken
GitHub Copilot bills hit $800 as Visual Studio adds usage alerts and MCP trust checks
Varonis "Rogue Agent" flaw exposes Google Dialogflow CX chatbots
DoorDash's DashBench shows paired and open models catch more bugs
Perplexity quietly builds an internal coding agent, "Teammate"
Gemini API expands managed agents with background tasks and remote MCP
AWS expands its DevOps Agent with AI-powered release management
Lilian Weng reframes recursive self-improvement around "harness engineering"
"HalluSquatting" turns AI coding assistants into botnet builders
Former GitHub CEO Thomas Dohmke unveils "Entire", a distributed Git network for AI agents
Martin Fowler's site: experiences with local models for coding
GitHub's AI coding agent leaks private repositories when asked nicely
A2UI v0.9 offers portable, framework-agnostic generative UI
Replit builds ViBench and Telescope for continual agent learning
Anthropic brings Claude Cowork to mobile and web as data shows most users aren't coding
AI coding erodes code maintainability, and a business springs up to fix it
"Agent orchestration is so two-years ago"
Block details Builderbot for orchestrating AI agents across the SDLC
Martin Fowler's team weighs the viability of local models for coding
The next DevOps bottleneck: managing more software than teams can handle
Import AI: OSWorld 2.0 raises the bar for computer-use agents
Import AI: JD weaves LLMs into industrial-scale inventory management
MCP adds centralised, enterprise-managed authorisation
Cloudflare and AWS embed x402 agent payments at the edge
SnortML brings machine learning into intrusion detection
"Why most AI projects fail: it's infrastructure and people"
Claude reaches GA on Microsoft Foundry, but European enterprises cannot deploy it
Vellum launches a memory-first personal assistant
"Microsoft, AWS and Anthropic are spending billions, and not on better models"
AIEWF closes with the loops debate over software factories
2026 AI Engineer Survey: agents move from drafting to doing
Vercel's Andrew Qu on why agents are a new kind of software
Microsoft launches $2.5B Frontier Company to embed AI engineers with customers
Devin Security Swarm scans whole codebases via Agentic MapReduce
Claude Enterprise gains admin analytics and spend controls
Smart model routing becomes a standard cost-control pattern
"Understanding is the new bottleneck" for AI-assisted coding
Trunk Tools cuts construction document review from 60 days to 10
Cloudflare's Skipper AI agent gives natural-language access to internal data
AIEWF Day 3: autoresearch and the tension between AI and human agency
The website of the future may assemble itself for every visitor
Skill engineering and the case against one-shot AI design
Continual Harness: an efficient self-improving agent on ARC-AGI-3
ZCode ships cross-platform as GLM-5.2's dedicated development environment
Meta quietly launches vibe-coded gaming app Pocket
Two-thirds of enterprises hedged their AI strategy after the Fable 5 outage
Alibaba's SkillWeaver cuts agent tool-routing token use by 99%
The "short leash" method for AI coding without losing control
Manufact launches an MCP cloud for shipping and monitoring MCP servers
Senior SWE-Bench aims to test agents as senior engineers, not juniors
How do you turn AI coding chaos into a repeatable playbook?
Lightrun adds pull request production-impact assessment
AIEWF Day 2: loops, software factories and forward deployed engineering
Forward Deployed Engineers and the future of software engineering
Claude Code deletes chat transcripts older than 30 days
Amazon launches $1 billion Forward Deployed Engineering organisation
X launches MCP server to expose platform data to AI tools
Anthropic removes covert steganographic code from Claude Code
Godot game engine bans vibe-coded contributions
Autoresearch: the feedback loop behind self-improving agents
Gemini Spark, Google's agentic assistant, arrives on Mac
Why your AI bill is bigger than it should be
Reliability comes from the system, not the agent
Anthropic adds Enterprise Gateway for Claude Code on AWS and Google Cloud
Restaurants can now accept orders placed directly from ChatGPT and Claude
GitLab 2026 report: AI accelerates coding but not overall software delivery
Devin Fusion: multi-model harness cuts AI coding costs 35%
Salesforce promotes Claude Tag in Slack despite competitive tension
Microsoft Copilot Autofix extends to Azure DevOps
AWS Lambda MicroVMs for hardware-isolated agent execution
AI agents identified as both a cause of database sprawl and a proposed remedy
Claude Code is steganographically marking its requests
Morgan Stanley: making agents less autonomous cut reconciliation risk in half
OpenAI announces Codex Micro Keyboard: dedicated hardware for AI coding
Mozilla: indirect prompt injections pose severe risk to AI coding agents
Harness adds autonomous AI agents to automate DevOps workflows
Elastic open-sources Atlas, an agent memory system based on cognitive science
OpenClaw launches on Android and iOS
AWS FinOps Agent: agentic cloud cost management in public preview
Notion kills Gmail client as AI agents handle majority of user inboxes
Amazon Q vulnerability: malicious Git repos can execute code and steal cloud credentials
ByteByteGo: structured memory is essential for long-horizon AI agents
Attackers exploit SimpleHelp flaw to steal data from AI coding assistants
Target builds an LLM-based system for marketing forecast matching
RAG vs Graph RAG vs Agentic RAG: Three Architectures Compared
Ford hired AI and sacked engineers. It backfired.
GPT-5.6 Sol/Terra/Luna Launches as Government-Restricted Preview
Botsitting: Workers Spend 6.4 Hours Per Week Supervising AI (Glean Work AI Index 2026)
Runtime verification is the missing layer for production coding agents
Claude Code turning every engineer into three — but teams now need product thinkers
OpenAI Internal Data: Codex Output Tokens Grew 56x in Research, 32x in Customer Support
Most companies think they're building a software factory. They're actually just shipping bugs faster
Grab Builds "Palana": Kubernetes-Native Secure Execution Platform for AI Agents
Cloudflare Ships Open-Source Agent Skills Library for Zero Trust Deployment
OpenAI Says Its Own Employees Are Moving from Chat to Agents
Vercel AI SDK 7: Zero-Overhead Execution Loop for Multi-Step Tool Calls
DeepReinforce Releases Ornith-1.0: State-of-the-Art Open-Source Coding Model Family
Meta Autodata: AI Agents That Build Better Training Datasets
Reward Hacking Benchmark: RL-Tuned Coding Agents Exploit Evaluation Flaws at 13.9%
Hugging Face Launches Single-Command vLLM Server Deployment on Serverless Jobs
OpenAI Kepler: Internal AI Agent Queries 600+ Petabytes via MCP and RAG
Stack Overflow for Agents: API-First Knowledge Exchange for AI Coding Agents
Dropbox Nova: Internal Platform for Orchestrating AI Coding Agents at Scale
Gemma 4 12B Enables On-Device Multimodal Agentic Workflows with Encoder-Free Architecture
Simon Willison: what happened after 2,000 people tried to hack his AI assistant
Agentic memory frameworks vary 28x in token cost for the same task
Google Expands Gemini Computer Use Capabilities
Informing LLMs About Existing Code Cuts Token Use and Reduces Security Issues
PostHog Rewrites SQL Parser to Be 70x Faster Using Parallel Claude Code Sessions
Language-Native Software: Natural Language Becoming the Deterministic Interface
Go Micro: Open Agent Harness with Built-In MCP and A2A Gateways
Hasp: Local-First Secret Broker with Time-Bounded Agent Access
Meta Releases Agent Harness Framework for Evaluation and Orchestration
Akrites: new coalition to protect open-source from AI-assisted supply chain attacks
AI Coding Agents Could Cost More Than the Developers Using Them
Database Vendors Pitch Smarter Data Retrieval to Contain Agentic AI Costs
Loop Engineering: Reducing Human Prompting Carries Hidden Costs
AWS Blocks: Open-Source TypeScript Framework for AI Agent Backends
AWS Lambda MicroVMs Extend to 8-Hour Runtime for Agent Workloads
OpenAI Codex Agent Bombards SSDs with Needless Write Operations
Anthropic Launches Claude Tag as Always-On Agentic Coworker in Slack
Prompt Debt: Natural Language Specifications Create Long-Term Technical Debt
Google Executive Predicts End of Direct Data Platform Access
Ex-Meta L8 Engineer Shares Agent-Agnostic Agentic Development Workflow
Claude Code Extended Thinking Output Is Encrypted and Not Verifiable as Authentic Reasoning
Microsoft Expands AKS with Bare Metal Nodes and AI Agent Infrastructure
DeepMind Publishes AI Control Roadmap for Securing Internal Agents
Vercel Launches Eve Open Source Agent Framework and Passport Shadow-AI Tool
Sakana Fugu: Multi-Agent Orchestration Exposed as a Single Model
Morph LLM Achieves 3x Codegen Speedup with Output-Trained Drafter
Loop Engineering: Agents Work Best in Iterative Evaluate-and-Reprompt Cycles
Google, Microsoft and Cisco Developing Agentic Resource Discovery Protocol
Agent Hooks Enforce Guardrails Deterministically Inside Agent Loops
Agent-Driven PR Pipeline Closes Issues Within Hours of Filing
ByteByteGo: Playbook for AI-Native Engineering Transformation at Scale
Cursor Acquires Continue, the Open-Source GitHub Copilot Alternative
Qodo Ships Cross-Repo Code Review for AI-Flooded Teams
Self-Harness: AI Agents Rewrite Their Own Execution Rules for Up to 60% Gains
Why Every DevOps Engineer Is Suddenly Learning MCP
Nvidia Research Shows Code Execution Outperforms Tool Calls for AI Spatial Reasoning
The Case for Rejecting AI Code That Works
OpenAI's Codex adds Record and Replay
Azure Functions ships a serverless agents runtime at Build 2026
GitLab 19.0 embeds agentic AI across the SDLC
CircleCI's Chunk Sidecars bring CI validation into the AI coding loop
Microsoft positions Windows to secure AI agents
Cloudflare Launches Temporary Accounts for AI Agents
Project Valkey Uses AI Bots to Automate Open-Source Backporting
Pinterest pivots search to taste with Ask Pinterest and an MCP layer
Builders race to replace Fable as Kimi K2.7-Code lands
Chainguard launches the Athena coalition to turn AI on open-source flaws
Claude Code adds shareable artifacts
Perplexity launches Brain, a self-improving memory for agents
Mistral adds Code and Apps to its Vibe assistant
GitHub Retires Its Free AI Model Playground
Stack Overflow Blog: AI Agents Need Contract-Based Capability Declarations
Checkmarx SAST Engine Focuses on Post-LLM Code Quality
Anthropic overhauls Claude Design to fix its token-burning problem
Vercel ships eve and Connect for production AI agents
Nvidia GEAR's agents learn to run their own robot labs
Study finds expertise, not coding skill, drives Claude Code results
AWS pushes continuous agentic DevOps with Kiro and Bedrock AgentCore
MCP Gets Its Missing Enterprise Authorization Layer
Cursor, GitLab and Zed Each Propose a Different Post-GitHub Future
New AI Optimization Framework Outperforms Claude Code and Codex by 2.5x
Adobe Brings Agentic Orchestration to the Full Creative Cloud Suite
GLM-5.2 detailed: the top open frontend-coding model, with IndexShare attention
GitHub ships a Copilot desktop app for parallel agentic workflows
Cursor teases a frontier model and an agent-native forge
DX research: AI-authored code hits 51.9% as pull requests balloon
Android 17 ships Android MCP and AppFunctions for on-device agents
AWS Launches Autonomous Agent for Continuous Codebase Modernization
Stack Overflow Blog: AI Agents Are a Confused Deputy with Access to Your Systems
SpaceX acquires Cursor for $60B in stock
Salesforce buys Fin for $3.6B to bolster Agentforce
Sakana launches Marlin, an autonomous research agent
Faros study quantifies the AI code-review burden
Anthropic pauses token-based billing for the Claude Agent SDK
Critical Microsoft Copilot "SearchLeak" flaw exposed 2FA codes and emails
Stanford's DeLM cuts multi-agent costs 50% without a central orchestrator
"Running local models is good now" tops Hacker News
Microsoft reportedly turns to AWS as GitHub hits an AI capacity crunch
Martin Fowler's team on building reliable agentic AI systems
OpenRouter Fusion blends models to rival a single frontier
Z.ai releases GLM-5.2 with usable 1M context
Cognition's FrontierCode sets a deliberately brutal coding benchmark
Google builds a Skills Marketplace for Gemini Enterprise
OpenAI and Anthropic diverge on agent context management
Anthropic explains how Claude builds its own execution harnesses
Tenet's "Agentjacking" turns Sentry errors into code execution
New Relic adds an open-source tool to observe AI coding
GitHub removes the PAT requirement for agentic workflows
"Vibe coding can build your pipeline; it can't explain it six months later"
NewCore raises $66M to give AI agents workplace identities
HashiCorp's Terraform MCP Server reaches general availability
SwirlAI argues against monitoring AI systems like web services
"MCP solved tool calling, A2A solved coordination, what solves transport?"
Pinecone connects AI agents to enterprise data via Microsoft OneLake
Azure Container Apps Sandboxes run untrusted agent code
WebMCP enters Chrome origin trials
Why AI retrieval and ranking need more than vector search
Coinbase gives AI agents wallet access
OutSystems bids to be the neutral agent orchestrator
OpenAI WebRTC audio session gains document context
An AI agent bankrupted its operator while scanning DN42
Moonshot's Kimi K2.7-Code arrives as an open coding model
Google ships a Colab CLI for developers and agents
Angular publishes official Agent Skills for AI coding tools
Cohere's North Mini Code lets developers stack their own AI
Xiaomi's open-source MiMo Code beats Claude Code on long tasks
Microsoft open-sources SkillOpt for tuning agent skills
GPT-5.5 and Codex reach general availability on Amazon Bedrock
Stack Overflow for Agents announced
Linear ships an autonomous coding agent
Google's GKE Agent Sandbox and Hypercluster
GitHub Copilot CLI reaches general availability
AWS MCP Server reaches GA; Agent Registry in preview
GitHub cuts agent token spend up to 62%
Slack on managing context in long-running multi-agent systems
An AI agent runs amok in Fedora and elsewhere
Agentic development hinges on verification, a runtime problem for cloud-native systems
GitLab previews a DevOps platform for the agentic era
Developers reportedly spend ~20 hours a week in Claude Code
Microsoft Foundry adds runtime and governance for production agents
A €0.01 transfer could compromise a banking AI agent
Azure API Management ships a unified model API and MCP content safety
Datadog extends AI-driven observability deeper into DevOps workflows
LinkedIn on AI as a new execution model for engineering
Claude Code adds dynamic workflows; Dropbox builds Nova
"Intent debt" as the new cost of agentic engineering
Anthropic says 80% of its new production code is authored by Claude
Apple approves the first AI agent for Messages Business Chat
Open Code Review released by Alibaba
Meta launches business AI agents across WhatsApp, Instagram and Messenger
Cloudflare and Stripe let agents deploy to production
OpenAI cuts agentic latency with a WebSocket execution mode
Microsoft launches Scout, an AI executive assistant
Karpathy's Autoresearch finds a three-year-old bug in PostHog
