Signal, not noise: five primary-source reads from September 21, each with a two-line note on why it matters for builders.
Refreshed repository pull requests page is generally available
The new repository pull requests page is now generally available to all GitHub users, with content assist for filters, AND/OR advanced search, a collapsible sidebar, compact presentation mode, and bulk actions added since the preview. Adopt the new filters and compact mode now so review triage across busy repositories costs minutes instead of context switches.
Read the full story at GitHub Changelog (September 21, 2026).
GitHub Enterprise adds credential inventory exports
Enterprise owners can now export a complete inventory of every credential that can reach their enterprise, covering SSH keys, classic and fine-grained personal access tokens, and OAuth App tokens, with a matching REST API for automation. Schedule a first export this week and wire it into incident-response runbooks so the next credential scare starts from evidence instead of guesswork.
Read the full story at GitHub Changelog (September 21, 2026).
Grok 4.7 is now available in GitHub Copilot
xAI's Grok 4.7 reasoning model is rolling out across Copilot surfaces from VS Code to the CLI, cloud agent, and JetBrains, aimed at agentic coding and complex multistep workflows under usage-based billing. Enable it behind your model policy for agent-heavy tasks first, then compare completion quality and credit burn against your current default before switching teams over.
Read the full story at GitHub Changelog (September 21, 2026).
OpenAI Academy expands with new learning paths
OpenAI Academy adds role-based pathways for developers building with Codex and the API, leaders driving adoption, educators, and college students, joining the existing Apply AI at Work track with assessed course badges. Point your developers at the Build with AI path and your leads at the AI Leadership course so experimentation turns into governed, repeatable delivery.
Read the full story at OpenAI (September 21, 2026).
Advisory group on mathematics and artificial intelligence
OpenAI convened an advisory group on mathematics and AI, signaling deeper investment in formal reasoning as a capability and evaluation frontier. If you build math-heavy copilots, tutors, or verification pipelines, track this group's output: it may set the next bar for rigorous model evaluation your customers will ask about.
Read the full story at OpenAI (September 21, 2026).