[juliusmarminge]

The server preview browser, chrome-headless-shell 154.0.8037.92, still trusts certificates from the compromised eMudhra intermediate CAs. Chrome blocks those keys through CRLSet, but chrome-headless-shell never loads a CRLSet. Electron 44.9.3 fixed the desktop app's own views in #22645, but that did nothing for the server browser.

Chromium has since added the eight keys to its compiled-in SPKI blocklist (crrev.com/c/8618876), which every chrome-headless-shell build carries. This PR pins 156.0.8078.70, the first M156 build with that change (crrev.com/c/8627687). The current Chrome for Testing stable, 156.0.8078.47, predates it. After upgrading, a server downloads the new browser the first time a preview tab opens.

156.0.8078.70 is in the Chrome for Testing bucket for all six platforms but is not yet listed in known-good-versions-with-downloads.json. Google has not yet promoted it to the Chrome Stable channel. Every zip matched the bucket's stored length and MD5 before I took its SHA-256.

Proof

No offline regression test is possible here. A chain that hits the blocklist needs one of the compromised CA keys. The proof is a runtime transcript through T3's own PreviewBrowser install and ServerBrowserContexts launch, using a persistent profile, against the disclosure site's test hosts.

Before (154.0.8037.92):

LOADED   https://dv-g3b-emudhra.we-have-a-problem.com/  "Your client successfully connected to the site. That's bad!"
LOADED   https://dv-g3c-emudhra.we-have-a-problem.com/  "Your client successfully connected to the site. That's bad!"
LOADED   https://ov-g3b-emudhra.we-have-a-problem.com/  "Your client successfully connected to the site. That's bad!"
LOADED   https://ov-g3c-emudhra.we-have-a-problem.com/  "Your client successfully connected to the site. That's bad!"
LOADED   https://ev-g3b-emudhra.we-have-a-problem.com/  "Your client successfully connected to the site. That's bad!"
LOADED   https://ev-g3c-emudhra.we-have-a-problem.com/  "Your client successfully connected to the site. That's bad!"
LOADED   https://dv-g3b-noct-emudhra.we-have-a-problem.com/  "Your client successfully connected to the site. That's bad!"
BLOCKED  https://*-aaa-emudhra.we-have-a-problem.com/ (7 hosts)  net::ERR_CERT_AUTHORITY_INVALID
LOADED   https://example.com/

After (156.0.8078.70, fresh T3 home, so this run also exercised the pinned linux64 download, size check and SHA-256 check):

BLOCKED  https://dv-g3b-emudhra.we-have-a-problem.com/  net::ERR_CERT_REVOKED
BLOCKED  https://dv-g3c-emudhra.we-have-a-problem.com/  net::ERR_CERT_REVOKED
BLOCKED  https://ov-g3b-emudhra.we-have-a-problem.com/  net::ERR_CERT_REVOKED
BLOCKED  https://ov-g3c-emudhra.we-have-a-problem.com/  net::ERR_CERT_REVOKED
BLOCKED  https://ev-g3b-emudhra.we-have-a-problem.com/  net::ERR_CERT_REVOKED
BLOCKED  https://ev-g3c-emudhra.we-have-a-problem.com/  net::ERR_CERT_REVOKED
BLOCKED  https://dv-g3b-noct-emudhra.we-have-a-problem.com/  net::ERR_CERT_REVOKED
BLOCKED  https://*-aaa-emudhra.we-have-a-problem.com/ (7 hosts)  net::ERR_CERT_AUTHORITY_INVALID
LOADED   https://example.com/

The ServerBrowser tests use a mock Chromium, so CI never runs the real browser. To cover the two-major jump, I sent every CDP method the server preview uses through Playwright 1.60.0 to both builds: screencast and frame acks, mouse and key input, Input.insertText, Fetch interception (continue, fulfill, fail), isolated worlds, screenshots, user-agent and metrics overrides, and cache and cookie clearing. Both builds accepted every call with identical results.

vp run check:changed passes.

Only linux64 was run end to end. The other five platforms' bytes and hashes come from the bucket downloads, which the installer checks on first use.

🤖 Generated with Claude Code (Claude Opus 5.5)

[macroscopeapp]

Approvability

Verdict: Needs Human Approval

Macroscope's review found this PR requires human approval — The server’s default headless Chrome release and integrity pins are being changed, causing new preview-browser installations to fetch and execute a different production binary. The change also alters certificate-trust behavior for compromised intermediates, making human review appropriate.

^{You can add or adjust custom eligibility rules. Learn more.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	5.0 KiB	5.0 KiB	0 B (0.0%)	6.8 KiB	✅
Codex	Thread snapshot wire	3.8 KiB	3.8 KiB	0 B (0.0%)	4.9 KiB	✅
Codex	Live turn WebSocket wire	1.2 KiB	1.2 KiB	0 B (0.0%)	2.0 KiB	✅
Codex	Live turn WebSocket decoded	20.9 KiB	20.9 KiB	0 B (0.0%)	29.3 KiB	✅
Codex	Live turn messages	2	2	0 (0.0%)	8	✅
Claude	Total thread wire	5.0 KiB	5.0 KiB	+31 B (+0.6%)	6.8 KiB	✅
Claude	Thread snapshot wire	3.8 KiB	3.8 KiB	0 B (0.0%)	4.9 KiB	✅
Claude	Live turn WebSocket wire	1.2 KiB	1.2 KiB	+31 B (+2.6%)	2.0 KiB	✅
Claude	Live turn WebSocket decoded	21.2 KiB	21.2 KiB	+41 B (+0.2%)	29.3 KiB	✅
Claude	Live turn messages	1	2	+1 (+100.0%)	8	✅

Baseline: be1d066 · PR result: 590a023 · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 108.5 KiB
• Claude decoded thread snapshot: 108.8 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.