fix(shared): reject corrupted T3 Connect authorization links

[shivamhwp]

Corrupted t3 connect links currently proceed through browser authorization and fail only when the CLI receives the code. Validate the state and PKCE challenge against the canonical base64url forms generated by the CLI before starting authorization. Both loopback and headless flows use this parser.

Closes #4934. The existing invalid-link page asks users to run t3 connect again.

Validation: 24 focused shared tests pass; both corruption regressions fail before the fix. Targeted lint, formatting and diff checks pass. Internal Fable refinement remains pending.

Stacked on #11298; this PR changes only the shared parser and its tests.

Model and harness: GPT-6 through Codex.

[macroscopeapp]

Approvability

Verdict: Not approved

Macroscope's review found this PR not approvable — The production change alters validation in the shared T3 Connect authorization flow, determining which browser links may proceed to Clerk authorization. Although the implementation and tests are narrowly scoped, authentication-sensitive behavior requires human review.

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

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

ℹ️ No successful main baseline artifact is available yet. This run establishes the initial measurement.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	—	13.6 KiB	—	15.1 KiB	✅
Codex	Thread snapshot wire	—	7.0 KiB	—	7.3 KiB	✅
Codex	Live turn WebSocket wire	—	6.5 KiB	—	7.8 KiB	✅
Codex	Live turn WebSocket decoded	—	57.1 KiB	—	66.4 KiB	✅
Codex	Live turn messages	—	10	—	21	✅
Claude	Total thread wire	—	13.6 KiB	—	15.1 KiB	✅
Claude	Thread snapshot wire	—	7.0 KiB	—	7.3 KiB	✅
Claude	Live turn WebSocket wire	—	6.5 KiB	—	7.8 KiB	✅
Claude	Live turn WebSocket decoded	—	57.8 KiB	—	66.4 KiB	✅
Claude	Live turn messages	—	9	—	21	✅

Baseline: unavailable · PR result: 4e60b59 · 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: 113.9 KiB
• Claude decoded thread snapshot: 114.6 KiB

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

[greatitself]

Independent validation of 4e60b590364e1c40417155de61a4b8e59b85b0f3 (merge-base 050e2cafb144).

Exercised the unused base64url bits systematically: 256 valid state/challenge pairs all round-tripped. For each pair, changed only unused bits in the final character: 15 state variants and 3 challenge variants. This revision rejected all 4,608 noncanonical variants; the merge-base accepted them. This is URL-parser validation, not an end-to-end authorization test.

Reproduction scripts and recorded results — first.mjs; the numbered metadata file pins this PR’s tested revisions.

Model: GPT-6. Harness: Codex.