FEATURED
06:21 UTC
Sealing the leaks in coding-agent evals: Cursor shows SWE-bench Pro scores are being gamed
data benchmark study
medium
Treat current coding-agent leaderboards as contaminated until you can reproduce results under a sealed, auditable eval harness.
openai
06:21 UTC
OpenAI previews GPT-5.6 (Sol/Terra/Luna) with new pricing and cache semantics under limited rollout
new product launch
high
GPT‑5.6 brings tiered models and new caching economics—rethink routing and budgets before you ship anything heavy.
openai
06:22 UTC
From chat to delegation: Codex data shows agents are becoming workflows, not answers
trend pattern
medium
Treat agents like services you delegate to and measure runs, handoffs, and skills—not chat messages.
anthropic
06:24 UTC
CI moves into the inner loop for AI agents
trend pattern
high
Treat verification as a first-class inner-loop concern or AI agents will turn your CI into a rework and cost machine.
github
06:25 UTC
Agent sessions are the new runtime: tools now let you orchestrate, isolate, and rate-limit them
trend pattern
high
Model compute, security, and observability around sessions—your agent platform will get faster, safer, and cheaper to operate.
claude-opus-46
06:26 UTC
Reco launches Agent Security to inventory and control AI agents across your enterprise stack
new product launch
medium
Agent security is shifting from blog-post theory to operational tooling—start with an inventory and shrink what your agents can touch before they act.
claude-code
06:27 UTC
SonarQube’s MCP server lands for Claude Code; 2.1.195 fixes risky tool matching
integration announcement
medium
Hook Claude Code to SonarQube via the new MCP server and upgrade to 2.1.195 to get safer tool routing and sturdier agent runs.
microsoft
06:29 UTC
Azure Migrate adds Copilot-powered code insights (preview) for AKS/App Service modernization
new feature deep dive
medium
Azure Migrate’s new Copilot-driven code insights turn repo scans into actionable AKS/App Service migration plans at scale.