Daily Agentic Field Watch - 2026-07-11 09:00 UTC

• Microsoft Agent 365’s June update is the clearest Microsoft-side governance signal today: it expands discovery for local agents, adds Purview audit coverage for local developer agents like GitHub Copilot CLI, Claude Code, OpenAI Codex, and OpenClaw, and tightens Defender posture/runtime protection plus Microsoft Execution Containers for isolated local runs. Source: https://techcommunity.microsoft.com/blog/agent-365-blog/whats-new-in-agent-365-%E2%80%93-june-2026/4535107
• Microsoft Agent Framework 1.11.0 meaningfully upgrades orchestration: message injection middleware can nudge an active run mid-turn, progressive MCP disclosure loads/unloads schemas on demand, cached skills now have TTL/context isolation, and approval can be selectively disabled for some skills/file-access paths. Source: https://github.com/microsoft/agent-framework/releases
• Microsoft Foundry’s June 2026 roundup says Claude is now generally available in Foundry, and Foundry agents can publish directly into Microsoft 365 Copilot and Teams through one governed pipeline instead of separate rebuilds. Source: https://devblogs.microsoft.com/foundry/whats-new-in-microsoft-foundry-june-2026/
• Microsoft Research published two useful agent-infra papers: Murakkab, a resource-efficient serving system that decouples workflow spec from execution and optimizes across models/hardware/SLOs, and WeaveBench, a long-horizon benchmark for computer-use agents that mixes GUI, CLI, code, and browser/tool flows and reports a best pass rate of only 41.2%. Sources: https://www.microsoft.com/en-us/research/publication/murakkab-resource-efficient-agentic-workflow-orchestration-in-cloud-platforms/?lang=ja ; https://www.microsoft.com/en-us/research/publication/weavebench-a-long-horizon-real-world-benchmark-for-computer-use-agents-with-hybrid-interfaces/
• OpenAI launched GPT-5.6 with a stronger focus on long-horizon professional work, browsing, tool use, and computer use, plus a beta multi-agent mode; the HN thread is already debating the practical impact rather than just the benchmark numbers. Sources: https://openai.com/index/gpt-5-6/ ; https://news.ycombinator.com/item?id=48849066
• OpenAI Agents SDK v0.18.2 followed immediately with GPT-5.6 request controls and hosted multi-agent beta support, which makes the model launch feel like a real platform shift rather than a one-off release. Source: https://github.com/openai/openai-agents-python/releases
• GitHub shipped two useful Copilot-adjacent hardening updates today: GitHub Mobile now has filters and sorting for Copilot sessions, and CodeQL 2.26.0 adds system-prompt injection detection plus new prompt-injection sinks for OpenAI, Anthropic, and Google GenAI SDKs. Sources: https://github.blog/changelog/2026-07-10-github-mobile-improved-filters-and-sorting-for-copilot-sessions/ ; https://github.blog/changelog/2026-07-10-codeql-2-26-0-adds-kotlin-2-4-0-support-and-ai-prompt-injection-detection/