Provenance Record

Clear Briefing of about 654 words (within the 300-800 range). Structure follows the changelog (rule, brownouts, options, scope), followed by a clearly separated Related Reporting section and an honest What We Don't Know list. Title is 132 characters (under 150).

Read all three snapshots from disk (source-0, source-1, source-2 .html.gz, all HTTP 200, no archive fallback, suspicious_patterns null for all). source-0 (GitHub changelog, Nov 2, 2026, status label 'Retired'): confirmed verbatim 'Starting December 2, 2026, GitHub Actions will block pull_request_target events by default in all public repositories, unless an applicable Actions event policy explicitly allows them. Enforcement was previously scheduled for November 2, 2026.'; the evaluate-mode sentence; brownouts Nov 9 and Nov 16, both 16:00-18:00 UTC; 'failed runs with an error annotation; the workflows won't execute' and 'return to evaluate mode, allowing workflows to run with warnings'; Insights; the two options (keep default rule / explicitly allow in an applicable Actions event policy, with workflow file targeting to allow specific workflows); the 'runs in the context of the base repository and can access its secrets' rationale; scope 'public repositories on GitHub Free, GitHub Team, and GitHub Enterprise Cloud, not GitHub Enterprise Server'; private and internal repositories not affected. The 'block by default' scope is public repositories only, as the article states. The changelog gives no reason for the move from Nov 2 to Dec 2, as the article says. source-1 (The Register, Jessica Lyons, Oct 29, 2026): confirmed 'identified at least 83 attacks against 61 public GitHub repositories' (the lead says 'Researchers identified at least 83 attacks...'; the body says Radware threat hunters, with Radware's Gabi Nakibly and Zvika Babo named), 'coordinated campaign to execute commands in CI/CD environments and steal credentials and access tokens', the event-triggers-workflow explanation, the quote 'incorporates attacker-controlled content directly into shell source code', and the least-privilege/separation recommendation. The snapshot contains zero occurrences of pull_request_target, so the article's statement that the report does not name it is correct and the article draws no link between the campaign and GitHub's change beyond saying so explicitly. Minor note: the article attributes the findings to 'researchers' without naming Radware, which is an omission, not an error. source-2 (StepSecurity, Varun Sharma, Oct 28, 2026): confirmed verbatim 'GitHub is introducing a native egress firewall for GitHub-hosted Actions runners' and 'GitHub's current firewall preview supports GitHub-hosted Linux runners'; the post compares the firewall with StepSecurity's own Harden-Runner, so the article's vendor-comparison label is accurate. The egress-firewall claim rests on StepSecurity alone and is worded as such; no GitHub announcement was cross-checked. Internal link to the Oct npm article checked: that article contains 'alongside the existing pull_request_target restriction'. Independent-source recount: 3 sources; The Register is independent of the subject (1); the GitHub changelog is the subject's own page; StepSecurity is a competing vendor and not independent for the egress-firewall claim. Independence is not required for a Briefing.

Every date, time, name, number and quote checked against the snapshots; no discrepancies. Quotes are verbatim (apostrophe style aside).

Accurate, well-hedged Briefing on a dated, actionable GitHub Actions policy change that affects maintainers of public repositories, with a time-sensitive first brownout on Nov 9. Approved. Featured candidacy: 1 - accurate, time-sensitive policy coverage for CI/CD maintainers, but a scheduled enforcement notice rather than a major event