AI Agent News: Releases, Launches and Usage Jumps, Live · page 3
-
dsh-Go gained 7,514 GitHub stars in a week
Now at 27.5K stars.
-
Graphify passes 150K GitHub stars
-
Headroom passes 80K GitHub stars
-
Inkeep downloads up 26% in a week
208K npm downloads in the 7 days to Nov 18, against 166K the week before.
-
Kimi Code downloads up 33% in a week
114K npm downloads in the 7 days to Nov 18, against 85.6K the week before.
-
Learnharness passes 10K GitHub stars
-
Mastra downloads up 27% in a week
4M npm downloads in the 7 days to Nov 18, against 3.2M the week before.
-
OpenShell downloads up 44% in a week
109K PyPI downloads in the 7 days to Nov 18, against 75.6K the week before.
-
Ruflo downloads down 30% in a week
66.1K npm downloads in the 7 days to Nov 18, against 95K the week before.
-
VoltAgent downloads up 43% in a week
55.6K npm downloads in the 7 days to Nov 18, against 38.9K the week before.
-
External sandboxes for coding agents
Simon Willison explores sandboxing approaches for coding agents to improve safety when running local models, in collaboration with Prime Radiant.
-
Harvard researchers find AI coding agents shifted the bottleneck to code review
Harvard researchers report that AI coding agents have shifted development bottlenecks from coding to code review.
-
CopilotKit 1.89.0 released
This release improves performance and reliability for long Intelligence threads, adds support for distinct agent display names, and expands MCP Apps with model-context sharing.
-
Govern and discover records with common metadata in AWS Agent Registry
AWS Agent Registry now supports custom metadata schemas that administrators can define to govern and discover agents, MCP servers, and skills with validated fields.
-
OpenAI’s GPT-6.2 Astra and GPT-6.2 Sol now available
OpenAI's GPT-6.2 Astra and GPT-6.2 Sol models are now available in GitHub Copilot, designed for agentic coding and code review with improved coding judgment.
-
Warp: Cloud agents failing to start for some customers
Major impact; resolved after 50 min.
-
Launch HN: StarSling (YC P25) – Self-improving CI for agent-volume repos
StarSling, a Y Combinator P25 company, launches self-improving CI for repositories with high agent volume.
-
Claude Design: Some users unable to sign in with Apple on claude.ai and the Claude iOS app
Major impact; resolved after 2 h 55 min.
-
Claude in Chrome: Some users unable to sign in with Apple on claude.ai and the Claude iOS app
Major impact; resolved after 2 h 55 min.
-
Build WhatsApp AI agents with Amazon Bedrock AgentCore
SMART, an open-source SDK and Terraform stack, enables deploying WhatsApp AI agents built with Amazon Bedrock AgentCore, handling webhooks, media, memory, and templating.
-
Advancing computer-use agents for healthcare teams
OpenAI announces advances in computer-use agents designed for healthcare teams.
-
Building the Agent Host Protocol for sessions that continue across devices
GitHub built an open Agent Host Protocol to maintain agent sessions across editors, apps, phones, and cloud environments.
-
GitHub Copilot Fusion updates: more availability, benchmarks, and models
GitHub Copilot Fusion expands with new models, broader availability across Copilot surfaces, and public leaderboard benchmarks.
-
GitHub Copilot app, Microsoft Ignite releases
GitHub Copilot app now supports Azure DevOps repositories, GitHub Enterprise Server instances, and browser access.
-
MAI Code 2 and MAI Code 2 Flash in GitHub Copilot
Microsoft AI Code 2 and Code 2 Flash models are now available in GitHub Copilot.
-
GitHub Copilot in JetBrains, Ignite releases
GitHub Copilot for JetBrains adds remote access, on-device voice dictation, browser use controls, and context usage shortcuts.
-
Monitor GitHub Copilot activity through Microsoft Agent 365
GitHub Copilot agent activity can now be streamed into Microsoft Agent 365 for centralized audit and control.
-
GitHub Copilot Evals is now in public preview
GitHub Copilot Evals is now in public preview, allowing users to optimize Copilot performance as prompts, models, and tools change.
-
Muse Is Now Available in Australia: The World’s First Personal AI Agent Built for Everyone
Meta's Muse personal AI agent launches in Australia with local connector integrations for creativity, shopping, property, fitness and entertainment.
-
Secret scanning on demand for developers and agents
GitHub secret scanning now runs on demand through Copilot, enabling developers and agents to catch secrets before code is merged.
-
Introducing GPT-6.2 Astra and Sol
OpenAI released GPT-6.2 Astra and Sol models with improved professional outputs, coding, computer use, and cost efficiency across ChatGPT Work, Codex, and API.
-
LoanPro's AI servicing assistant cuts account summaries from 3–5 minutes to under 15 seconds
In an AWS blog post co-written with LoanPro, the lending platform describes a multi-agent assistant built with AllCloud. It runs on Amazon Bedrock AgentCore and the Strands Agents SDK and helps servicing agents get up to speed on a borrower's account during live calls.
-
How LoanPro built an agentic AI loan servicing assistant with Amazon Bedrock AgentCore
LoanPro built a multi-agent AI assistant on Amazon Bedrock to give loan servicing agents borrower context and recommendations, reducing call times by up to 15 percent.
-
In 27,280 coding agent runs, 64% of inference spend went to runs that failed
Analysis of 27,280 coding agent runs found that 64% of inference spending occurred on runs that ultimately failed.
-
Langfuse 4.82.0 released
feat(cloud): gate legacy v3 surfaces behind cutoff switches; feat(api-keys): add Developer preset and rename Ingestion to Application; feat(native): normalize provider usage and supplied costs
-
Qwen Code 0.28.0 released
feat(web-shell): add sidebar workspace filter; feat(managed-agent): add the DingTalk adapter on the managed channel path; fix(managed-agent): settle stale public-child recovery as terminal instead of retrying forever (+2 more releases that day)
-
Skyvern: Browser sessions and agents failing to start
Major impact; resolved after 12 h 5 min.
-
Fin outbound email delays lasted 21 h 20 min, 28.6x the typical incident
Fin, an AI customer agent for support conversations, had a major incident with delays in outbound email delivery. According to the official status page, it ran from Nov 12, 21:17 UTC to Nov 13, 18:36 UTC.
-
Corsair reaches 53.9K weekly downloads, up 70% on its 4-week average
Corsair, a tool for connecting your users to their apps, logged 53.9K downloads on npm in the week ending Nov 15, 2026. The previous weekly record was 33.3K.
-
Claude Cowork's Microsoft 365 connector errors lasted 27 h 57 min
Claude Cowork had elevated errors on its Microsoft 365 connector for 27 h 57 min, according to the official status page. The incident was rated major and ran 29.2x longer than the product's typical incident.
This list shows notable items; the sections above show everything of their kind, including patch releases of smaller projects and low-impact incidents. Summaries of outside articles are written automatically from the source's own text and may miss nuance, so follow the link before relying on one.