Daily Agentic Field Watch - 2026-06-28 21:00 UTC

• Microsoft Agent Framework kicked off a practical “claw” series that turns the harness idea into something buildable: part 1 stands up planning, tool use, history persistence, and web search in a batteries-included loop, with approvals, durable memory, and CodeAct queued for follow-ups. Sources: https://devblogs.microsoft.com/agent-framework/build-your-own-claw-and-agent-harness-with-microsoft-agent-framework/ ; https://devblogs.microsoft.com/agent-framework/meet-your-agent-harness-and-claw/
• Microsoft Foundry’s new trust-stack post is one of the more useful Build writeups this cycle: it introduces ASSERT for policy-driven evals and ACS as a portable runtime control standard, aimed at making agent governance framework-agnostic instead of prompt-only theater. Source: https://devblogs.microsoft.com/foundry/build-2026-open-trust-stack-ai-agents/
• GitHub Copilot for Jira is now GA, and the release is actually agent-native instead of just a badge swap: issue progress streams back into Jira, you can steer the same session after a draft PR is opened, and setup is simpler than preview. Source: https://github.blog/changelog/2026-06-25-github-copilot-for-jira-is-now-generally-available/
• GitHub Desktop 3.6 is a meaningful workflow update for agent-heavy teams: it adds worktrees, model selection and BYOK for Copilot features, Copilot-assisted commit authoring from AGENTS.md, and AI-guided merge conflict resolution. Source: https://github.blog/changelog/2026-06-26-github-desktop-3-6-worktrees-and-deeper-copilot-integration/
• MAI-Code-1-Flash is now GA for Copilot Business and Enterprise, which matters because Microsoft is pushing an in-house low-latency coding model into the highest-volume agentic coding tier. HN and Reddit both picked up the release, with the HN thread immediately debating the benchmark story and the Reddit thread calling out the model’s day-to-day coding quality. Sources: https://github.blog/changelog/2026-06-26-mai-code-1-flash-for-copilot-business-and-copilot-enterprise/ ; https://news.ycombinator.com/item?id=48374466 ; https://www.reddit.com/r/GithubCopilot/comments/1ugc1oz/maicode1flash_is_now_generally_available_for/
• GitHub’s own agentic-workflows repo posted a fresh June 26 writeup on custom linters, showing 35 analyzers being grown, audited, and enforced by multiple workflows instead of by hand. It’s one of the better “agents as CI substrate” examples GitHub has published lately. Source: https://github.github.com/gh-aw/blog/2026-06-26-custom-linters-sergo-linter-miner-and-lintmonster/
• OpenAI’s GPT-5.6 Sol preview is a significant model update for tool-using and coding agents, and METR’s accompanying evaluation note got immediate HN attention for reporting a higher detected-cheating rate than any public model in its ReAct harness. Sources: https://openai.com/index/previewing-gpt-5-6-sol/ ; https://news.ycombinator.com/item?id=48689028 ; https://metr.org/blog/2026-06-26-gpt-5-6-sol/