Sanitize upstream titles in release-readiness tracker tables (embedded newlines + unescaped pipes)

[PureWeen]

Note

Are you waiting for the changes in this PR to be merged?
It would be very helpful if you could test the resulting artifacts from this PR and let us know in a comment if this change resolves your issue. Thank you!

Problem

The [Release Readiness] tracker issues render markdown tables built from upstream-controlled content — issue/PR titles dropped into pipe-delimited rows (| #link | <title> | <…> |). Several kinds of malformed (or hostile) title corrupt those rows:

1. Embedded newlines. Some ci-scan issues have a title containing a literal newline — observed live on issue [ci-scan] Recurring: MemoryLeakB42329 and Issue1939Test fail with Appium timeout in Controls ListView on macOS MacCatalyst (maui-pr-uitest [Content truncated due to length] #35957, whose real GitHub title spans two physical lines and ends with (maui-pr-uitest\n[Content truncated due to length]. An embedded \r\n splits the row across two physical lines, so the title tail + trailing cells land on a line that no longer contains the issue link. This was visible on the live trackers: net10-sr9 [Release Readiness] .NET 10 SR9 — shipped (release/10.0.1xx-sr9) #35867 (2 broken rows) and net11-preview6 [Release Readiness] .NET 11.0 preview6 — release/11.0.1xx-preview6 #35866 (3 broken rows).
2. Unescaped pipes. A literal | in a title (common — e.g. Fix A | B) injects an extra column. The preview engine already escaped pipes everywhere via its shared Format-MarkdownCell, but on the SR side cells were escaped ad-hoc — several embedded titles with no newline collapse at all.
3. Escape-the-escaper pipe breakout. A title may legally contain a literal \| (backslash immediately followed by a pipe). Escaping only the pipe turns that into \\|, which GitHub-flavored Markdown renders as a literal \ followed by an active column delimiter — so the row still breaks out. Both engines had this latent bug.
4. Raw </> (HTML injection / display loss) on the SR side. SR dropped titles in raw, so a title like Crash <!-- could inject an HTML-comment opener, and a legitimate List<T> (or engine-authored placeholder text like Bump <PatchVersion> …) was silently swallowed by GitHub as an unknown HTML tag and never displayed.

The ultimate root cause of (1) is upstream (the CI Failure Scanner producing a multi-line title), but a readiness engine should defensively sanitize any external content it embeds into its own tables.

Fix

• Get-ReleaseReadiness.ps1 — introduce a single null-safe Format-MarkdownTableCell helper and route every SR site that embeds upstream-controlled text through it. This covers the ci-scan rows, the Open PRs Targeting <srBranch>, regression classification, 🔴 Blocking summary, 🧹 Cleanup, 📥 Open Fix PRs Inbound, and Ship-readiness checks tables, plus the candidate-PR bulleted list. The helper:
  • collapses [\r\n]+ → space (hazard 1),
  • escapes each | → \| and doubles only the backslash run immediately preceding that pipe (via a single (\\*)\| regex pass), so a pre-existing \| becomes \\\| (renders a literal \|, no breakout — hazards 2 & 3). The doubling is scoped to pipe-adjacent runs rather than every backslash, so a title's other backslash escapes (\[link\](url), \*not emphasis\*) are preserved verbatim and not de-escaped into active Markdown. No-pipe-adjacent-backslash titles are unchanged (a | b → a \| b).
  • escapes </> → </> (hazard 4), matching the preview engine for SR↔Preview parity.
• Get-PreviewReadiness.ps1 (Format-MarkdownCell) — the newline collapse plus the same pipe-adjacent backslash handling so the preview engine is immune to the \| breakout too.

On </> escaping (now consistent across both engines): escaping angle brackets to entities has zero visual cost — <T> renders as <T> — so List<T> fidelity is preserved while raw HTML injection is neutralized (<!-- → <!--). It also fixes a latent display bug: engine-authored NextAction text such as Bump <PatchVersion> in eng/Versions.props was previously rendered raw and swallowed by GitHub as an unknown HTML tag, so the Release Captain saw Bump in eng/Versions.props; it now displays correctly. SR is additionally hash-freeze-immune (it emits its own hash at the top of the body, extracted with head -n1) and its human-notes markers are matched full-line-anchored, so escaping <> is defense-in-depth layered on top of those backend invariants rather than the sole protection. (.Trim() only touches leading/trailing whitespace.)

Tests

Deterministic, offline assertions:

• Format-MarkdownTableCell / Format-MarkdownCell unit tests — pipe escaping, LF/CRLF-run collapse, newline+pipe together, null/empty → empty string, whitespace trim; angle brackets escaped to </> (both engines, parity); literal \| does NOT break out (→ A \\\| B); non-pipe backslash preserved (C:\dir unchanged) and author-escaped non-pipe Markdown not de-escaped (\[link\](url) unchanged) for both engines; <!-- opener neutralized to <!--.
• SR ci-scan row — an embedded-newline title renders as a single physical row with its tail + age intact.
• SR tables (end-to-end Format-MarkdownReport) — a piped+newline title in the Open PRs Targeting, regression classification, 🔴 Blocking summary, 📥 Open Fix PRs Inbound, and Ship-readiness checks tables each stays on one physical row, pipe escaped, trailing column intact; the BLOCKED ship-check next-action with <PatchVersion> renders entity-escaped (so GitHub actually displays it).
• Human-notes marker-forgery regressions (security) — a title embedding …\n<!-- …:human-notes:begin -->\n… in a table cell and in the candidate-PR list must leave exactly one anchored begin-marker in the rendered body (the legitimate one), proving a hostile title cannot forge a second notes region.
• Preview engine — Format-MarkdownCell collapses LF/CRLF runs and preserves the existing pipe / angle-bracket escaping contract.

The discriminating assertions were verified red on the pre-fix scripts and green after (and the surgical-scoping assertions were verified red against the earlier global-doubling commit). Offline suite: 566 passed / 0 failed; full E2E: 632 / 0.

Scope / follow-ups (intentionally out of this PR)

• Backtick (inline-code) is intentionally not escaped: doing so would degrade the very common legitimate case of code-quoted titles like `CollectionView`, and an unescaped backtick is a cosmetic-only, non-structural concern (it cannot create a new column, inject HTML, or forge a human-notes marker, all of which require |/<, which are escaped).
• Null-safety of .title.Length under Set-StrictMode -Version Latest is pre-existing (titles are non-null by GitHub API contract) and intentionally deferred to a focused follow-up rather than mixed into this rendering PR.
• The SR "Reverts" table's "Reverts commit" column shows ? for every row (the This reverts commit <sha> body-regex never resolves). Pre-existing and unrelated; noted for a future pass.
• Filing an upstream issue against the CI Failure Scanner for the malformed (multi-line) titles is worth doing separately so the trackers receive clean input at the source.

[github-actions]

🚀 Dogfood this PR with:

⚠️ WARNING: Do not do this without first carefully reviewing the code of this PR to satisfy yourself it is safe.

curl -fsSL https://raw.githubusercontent.com/dotnet/maui/main/eng/scripts/get-maui-pr.sh | bash -s -- 36031

Or

• Run remotely in PowerShell:

iex "& { $(irm https://raw.githubusercontent.com/dotnet/maui/main/eng/scripts/get-maui-pr.ps1) } 36031"

[github-actions]

Skill Validation Results

@PureWeen — new skill validation results are available based on this last commit: 4f7d879.
To request a fresh validation after new comments or commits, comment /evaluate-skills.

✅ Skill Validation Results — 4f7d879 · Sanitize upstream titles in release-readiness tracker tables (embedded newlines + unescaped pipes) · 2026-06-20T20:43:44Z

✅ Static Checks Passed

Skills: 20 | Eval specs linted: 7

Full lint output

── .github/skills/agentic-labeler/tests/eval.vally.yaml
npm warn deprecated prebuild-install@7.1.3: No longer maintained. Please contact the author of the relevant native addon; alternatives are available.
✔ .github/skills/agentic-labeler/tests/eval.vally.yaml is valid
── .github/skills/code-review/tests/eval.capability.vally.yaml
✔ .github/skills/code-review/tests/eval.capability.vally.yaml is valid
── .github/skills/code-review/tests/eval.vally.yaml
✔ .github/skills/code-review/tests/eval.vally.yaml is valid
── .github/skills/code-review/tests/hermeticity.vally.yaml
✔ .github/skills/code-review/tests/hermeticity.vally.yaml is valid
── .github/skills/evaluate-pr-tests/tests/eval.vally.yaml
✔ .github/skills/evaluate-pr-tests/tests/eval.vally.yaml is valid
── .github/skills/try-fix/tests/eval.vally.yaml
✔ .github/skills/try-fix/tests/eval.vally.yaml is valid
── .github/skills/verify-tests-fail-without-fix/tests/eval.vally.yaml
✔ .github/skills/verify-tests-fail-without-fix/tests/eval.vally.yaml is valid

⏭️ LLM Evaluation: Skipped

No changed skills with eval specs found.

🔍 Full results and investigation steps

[MauiBot]

AI Review Summary

@PureWeen — new AI review results are available based on this last commit: eb765c0. To request a fresh review after new comments or commits, comment /review rerun.

Review Sessions — click to expand

Gate — Test Before & After Fix

Gate Result: ⚠️ SKIPPED

No tests were detected in this PR.

Recommendation: Add tests to verify the fix using the write-tests-agent.

---

Pre-Flight — Context & Validation

Issue: #36031 - PR-local infrastructure bug report (no linked issue detected)
PR: #36031 - Harden release-readiness tracker tables against embedded newlines in upstream ci-scan titles
Platforms Affected: windows; infrastructure/release-readiness markdown generation
Files Changed: 2 implementation, 1 test

Key Findings

• The PR changes .github/skills/release-readiness/scripts/Get-ReleaseReadiness.ps1, .github/skills/release-readiness/scripts/Get-PreviewReadiness.ps1, and .github/skills/release-readiness/tests/Test-ReleaseReadiness.ps1.
• Root cause: upstream ci-scan issue titles can contain literal CR/LF, which split generated markdown table rows across physical lines.
• Gate was already skipped by the caller because no tests were detected by the gate phase; local candidate testing used the release-readiness offline test script.
• Local baseline offline suite exits nonzero before alternatives because two notes-marker assertions fail on Windows line endings; candidate comparisons treat those as baseline unless a candidate explicitly changes that area.

Code Review Summary

Verdict: LGTM
Confidence: high
Errors: 0 | Warnings: 0 | Suggestions: 2

Key code review findings:

• 💡 Add a symmetric SR CRLF assertion for Format-CiScanIssueRows if desired.
• 💡 SR table-cell centralization is a reasonable future refactor, but not required for the observed ci-scan title bug.

Fix Candidates

#	Source	Approach	Test Result	Files Changed	Notes
PR	PR #36031	Collapse [\r\n]+ in Preview Format-MarkdownCell and SR ci-scan title before table rendering	⚠️ Gate skipped by caller; local relevant regression assertions pass, full suite has unrelated baseline failures	.github/skills/release-readiness/scripts/Get-PreviewReadiness.ps1, .github/skills/release-readiness/scripts/Get-ReleaseReadiness.ps1, .github/skills/release-readiness/tests/Test-ReleaseReadiness.ps1	Original PR

---

Code Review — Deep Analysis

Code Review — PR #36031

Independent Assessment

What this changes: Adds CR/LF collapse before markdown table output can consume upstream ci-scan issue titles, preventing embedded newlines from splitting tracker table rows. Preview applies this in the shared Format-MarkdownCell; SR applies it to the observed ci-scan title cell.
Inferred motivation: A malformed upstream ci-scan title containing a literal newline broke release-readiness tracker markdown tables.

Reconciliation with PR Narrative

Author claims: The PR hardens SR and Preview readiness tracker tables against embedded newlines and adds deterministic offline tests.
Agreement/disagreement: The code matches the narrative. The SR fix is intentionally targeted; Preview uses the broader shared formatter.

Prior Review Reconciliation

No prior ❌ Error findings found. GitHub CLI auth was unavailable; public PR metadata/comments and local branch diff were used.

Blast Radius Assessment

• Runs for all instances: No — only release-readiness report-generation scripts.
• Startup impact: No.
• Static/shared state: No.

CI Status

• Required-check result: undetermined locally because gh is unauthenticated.
• Classification: public PR metadata shows skill/static validation success; local release-readiness offline suite has two baseline notes-marker failures unrelated to this PR before candidate changes.
• Action taken: confidence capped for CI auth limitation in local review context.

Findings

💡 Suggestion — Add symmetric SR CRLF assertion

The SR ci-scan regression test uses LF; the implementation handles CRLF via [ ]+. A CRLF-specific SR assertion would improve symmetry with Preview tests.

💡 Suggestion — SR table-cell centralization remains future work

SR still has scattered table-cell escaping in other sections. The PR documents centralization as future work; this is not a blocker for the observed ci-scan title bug.

Failure-Mode Probing

• Malformed ci-scan title with LF: remains one physical markdown row.
• CRLF run in Preview Format-MarkdownCell: collapses to one space.
• Null Preview cell: existing guard returns empty string.
• Title with newline and pipe: newline collapse precedes pipe escaping.

Verdict: LGTM

Confidence: high for code correctness; local CI status collection limited by missing gh authentication.
Summary: The PR fix is minimal, correctly ordered, and covered by deterministic tests. No blocker was found in the PR fix itself.

---

Fix — Analysis & Comparison

Fix Candidates

#	Source	Approach	Test Result	Files Changed	Notes
1	try-fix	Broad SR Format-MarkdownCell helper used across SR table cells; Preview unchanged from PR	⚠️ 601 passed / 2 failed (same unrelated baseline notes-marker failures)	1 file	More robust SR coverage, but larger blast radius and helper-name collision risk in dot-sourced tests
2	try-fix	Row-boundary CR/LF collapse in ci-scan row appenders; Preview helper reverted and test moved to Add-CiScanTable	⚠️ 600 passed / 2 failed after fixing a PowerShell parsing error (same unrelated baseline failures)	3 files	Narrowest diff, but less robust than PR because other Preview table cells lose newline protection
3	try-fix	Broad SR helper plus LF-only release-captain notes block normalization; Preview unchanged from PR	✅ 603 passed / 0 failed	1 file	Full suite green, but only because it also fixes an unrelated notes-marker Windows line-ending baseline bug
PR	PR #36031	Targeted SR ci-scan title collapse plus shared Preview Format-MarkdownCell collapse	⚠️ Gate skipped; local relevant tests pass, full suite has unrelated baseline failures	3 files	Best scoped fix for the PR's stated issue

Cross-Pollination

Model	Round	New Ideas?	Details
maui-expert-reviewer	1	Yes	Candidate 1: centralize SR table-cell sanitation with an SR helper
maui-expert-reviewer	2	Yes	Candidate 2: sanitize at ci-scan row emission boundary to avoid broad behavior changes
maui-expert-reviewer	3	Yes	Candidate 3: combine broad SR helper with unrelated notes-marker line-ending baseline fix

Exhausted: Yes
Selected Fix: PR's fix — it is the best scoped fix for the embedded ci-scan newline problem. Candidate 3 is the only all-green local run, but it wins by adding an unrelated baseline repair; Candidate 1 is broader than necessary; Candidate 2 is narrower but loses the PR's broader Preview formatter hardening.

---

Recommended PR Title & Description

Assessment: ✏️ Recommend updating — the current title is good, but the description overstates the Windows test outcome available in the PR-agent artifacts.

Recommended title

Release-readiness: Harden tracker tables against embedded ci-scan title newlines

Recommended description

### Problem

The `[Release Readiness]` tracker issues render a **"Recent CI Failure Scanner signals"** markdown table built from upstream `ci-scan` issues. Some of those issues have a title that contains a **literal newline** — observed live on issue #35957, whose real GitHub title spans two physical lines and ends with `(maui-pr-uitest\n[Content truncated due to length]`.

Both readiness engines copy the raw title straight into a table cell (`| #link | <title> | <age> |`). They escaped the pipe character (and the preview engine also escapes `<`/`>`) but **neither stripped newlines**. An embedded `\r\n` splits the table row across two physical lines, so the row's title tail + age cell land on a line that no longer contains the issue link — breaking the rendered markdown table.

This was visible on the live trackers after the most recent run: **net10-sr9 #35867** (2 broken rows) and **net11-preview6 #35866** (3 broken rows). net10-sr8 #35876 happened not to contain such a title and rendered fine.

The ultimate root cause is upstream (the CI Failure Scanner producing a multi-line title), but a readiness engine should defensively sanitize external content it embeds into its own tables — exactly as it already does for `|` and `<`/`>`.

### Fix

- **`Get-ReleaseReadiness.ps1` (`Format-CiScanIssueRows`)** — collapse `[\r\n]+` → single space before escaping the ci-scan title cell.
- **`Get-PreviewReadiness.ps1` (`Format-MarkdownCell`)** — the same collapse, applied in the shared cell formatter so **every** preview table cell is protected (not just ci-scan).

`.Trim()` only removes leading/trailing whitespace, so it never handled the internal newline. Normal GitHub issue titles cannot contain newlines, so this only affects the malformed ci-scan titles.

### Tests

Adds 7 deterministic, offline assertions:

- **SR engine:** a ci-scan issue whose title contains an embedded newline now renders as a **single** physical table row, with the title tail + age still on that row.
- **Preview engine:** `Format-MarkdownCell` collapses LF and CRLF-runs to a single space, leaves no CR/LF in the cell, and **preserves the existing pipe / angle-bracket escaping contract**.

The discriminating newline assertions were verified to fail on the pre-fix code and pass after the fix. On the available Windows PR-agent run, the relevant regression assertions pass; the full offline suite still has unrelated pre-existing notes-marker baseline failures.

### Scope / follow-ups (intentionally out of this PR)

- The SR engine escapes table cells inline at multiple sites; only the ci-scan title site manifests the newline bug today (it's the only cell fed unsanitized upstream titles), so only that site is hardened here. Centralizing SR cell escaping (to match the preview engine's `Format-MarkdownCell`, including `<`/`>` parity) is a larger refactor left as future work.
- The SR "Reverts" table's "Reverts commit" column shows `?` for every row (the `This reverts commit <sha>` body-regex never resolves). Pre-existing and unrelated to this change; noted for a future pass.
- Filing an upstream issue against the CI Failure Scanner for the malformed (multi-line) titles is worth doing separately so the trackers receive clean input at the source.

---

Report — Final Recommendation

Comparative Fix Report - PR #36031

Candidates compared

Rank	Candidate	Result	Assessment
1	pr	Relevant regression assertions pass; gate skipped; full local Windows suite has unrelated notes-marker baseline failures	Best scoped fix. It hardens the known SR ci-scan title path and the shared Preview markdown cell formatter without pulling unrelated baseline repairs into the PR.
1	pr-plus-reviewer	Same as pr	Expert reviewer produced no actionable findings, so this candidate is identical to pr.
3	try-fix-3	Full local suite passed (603 passed / 0 failed)	Technically green, but wins by also fixing an unrelated Windows notes-marker baseline issue. That broader unrelated change is not the right fix for this PR.
4	try-fix-1	Relevant newline assertions passed; full suite still had the same unrelated baseline failures	More comprehensive SR cell centralization, but larger blast radius and introduces helper-name collision risk when scripts are dot-sourced together.
5	try-fix-2	Relevant SR/Preview ci-scan row assertions passed; full suite still had the same unrelated baseline failures	Narrow and avoids helper collision risk, but weakens the PR by removing Preview's shared formatter hardening for non-ci-scan table cells.

Key comparison

try-fix-3 is the only candidate with a fully green local suite, but its extra green result comes from an unrelated notes-marker line-ending repair. The requested bug is table breakage caused by embedded newlines in upstream ci-scan titles; the PR fix addresses that directly and includes targeted tests without changing unrelated tracker behavior.

try-fix-1 is a plausible future refactor for SR table sanitation, but it changes more SR table emitters than needed for the observed bug. try-fix-2 is narrower, but less robust than the PR because it protects only Preview ci-scan row emission instead of preserving the PR's shared Format-MarkdownCell newline defense.

Winning candidate

Winner: pr

The raw PR fix is the best candidate because it is narrowly scoped to the real bug, preserves existing escaping contracts, adds deterministic regression coverage, and avoids unrelated baseline repairs or broader SR refactors. pr-plus-reviewer is equivalent because the expert reviewer had no actionable feedback.

---

Future Action — review latest findings

No alternative fix was selected for this run. Review the session findings and CI results before merging.

[kubaflo]

Note

🤖 This review was automatically generated by a multi-model AI review system (Claude Opus 4.8, GPT-5.5, Gemini 3.1 Pro). Three models independently reviewed the code, then cross-pollinated their findings to produce this consolidated review.

Multi-Model Review — Round 1

Verdict: LGTM

After independent review and cross-pollination, all three models converged on LGTM:

• Gemini: LGTM
• GPT: NEEDS_DISCUSSION → LGTM (retracted CI concern after cross-pollination)
• Opus: LGTM

---

Summary

This PR adds robust newline handling to Format-MarkdownCell in the release-readiness tracker suite, preventing embedded newlines in upstream CI scan titles from corrupting Markdown table rows. The fix is correct, safe, and well-tested.

Key validation (Opus empirical verification):

• ✅ Regex [\r\n]+ is safe (no ReDoS), correct PowerShell semantics
• ✅ All 5 PR assertions pass post-fix (leading/trailing trim, CRLF collapse, $null safety)
• ✅ Tests are discriminating — fail pre-fix, pass post-fix
• ✅ No bugs in the changed code

---

Non-Blocking Suggestions

💡 SR cell escaping consolidation

Location: Test-ReleaseReadiness.ps1:2848

The SR ci-scan cell escapes | and newlines but not </>. Preview's Format-MarkdownCell escapes all four. Consider consolidating to a shared escaping helper (already noted in PR description as deferred follow-up: "centralize SR escaping").

Sibling sites (L3196/L3386/L3439) also embed PR/issue titles with only pipe escaping — same pre-existing pattern.

💡 Test clarity

Location: Test-ReleaseReadiness.ps1:2276

The first SR assertion is non-discriminating (would pass even without the fix). The second assertion is the real regression guard. Consider adding a clarifying comment.

---

What All Models Agree On

✅ Fix is correct: [\r\n]+ → single space handles all edge cases
✅ No injection risk: Regex is safe, orthogonal to existing pipe/angle escaping
✅ Tests validate the fix: Regression tests discriminate pre/post-fix behavior
✅ CI concern resolved: maui-pr: skipping is expected path-filtering for .github/skills/** changes

---

Confidence Assessment

High — All three models converged to LGTM with no blocking findings. Opus performed empirical PowerShell 7.5.4 validation. The fix is narrow, safe, and well-tested.

---

[kubaflo]

💡 💡 Minor (test clarity, non-blocking). This first assertion (issue row is a single physical line → Expected 1) is non-discriminating: I verified it passes on the pre-fix code too. With the embedded newline, the pre-fix row splits as | [#35957](...) | ...tail / [Content truncated...] | age |, and only the first physical line contains #35957, so the count is still 1. The real guard against the regression is the second assertion (title tail + age stay on that row), which I confirmed fails ❌ on pre-fix and passes ✅ post-fix. Consider a brief comment noting the second assertion is the discriminating one, so a future maintainer doesn't weaken it assuming the first already covers row integrity. (Gemini and GPT both concurred with this in cross-review.)

[kubaflo]

💡 💡 Minor (test clarity, non-blocking): I agree with Opus that the first new SR assertion (issue row is a single physical line) is not the discriminating regression check by itself: pre-fix output can still produce exactly one physical line containing #35957 while the title tail and age are pushed to the next line. The following assertion (title tail + age stay on that row) is the one that actually catches the bug. Consider a brief comment so future maintainers do not weaken the second assertion thinking the first one fully covers row integrity.

[kubaflo]

💡 💡 Suggestion (Test clarity): As noted by Opus, the first assertion (issue row is a single physical line -> Expected 1) is non-discriminating and passes on the pre-fix code as well, because the embedded newline splits the row such that only the first physical line contains the #35957 issue number. The real guard against the bug is the second assertion (title tail + age stay on that row). Consider a brief comment noting that the second assertion is the discriminating one.

[kubaflo]

💡 💡 Suggestion (non-blocking): Consider also escaping < and > to < and > here to match the escaping done in Get-PreviewReadiness.ps1's Format-MarkdownCell. This would prevent HTML injection if a ci-scan issue title contains angle brackets (e.g. List<T>). As noted in the PR description, centralizing cell escaping is a planned follow-up, which would be the perfect time to address this and other pre-existing unescaped sites (like L3196, L3386, and L3439) that Opus identified.

[kubaflo]

💡 💡 Suggestion (non-blocking): The newline collapse here looks correct, but the SR tracker still handles markdown-cell escaping ad hoc. Gemini's angle-bracket point is valid as a consistency/content-fidelity concern (Format-MarkdownCell escapes </> in the Preview path), and Opus also found sibling SR rows that embed upstream titles without pipe escaping or newline collapse. This PR is fine as-is for the observed ci-scan newline bug, but the planned follow-up to centralize SR cell formatting should cover this ci-scan title plus the other upstream-title sites, and should consider matching Preview's pipe/newline/angle-bracket escaping contract.

[kubaflo]

💡 💡 Suggestion (non-blocking, pre-existing scope). The newline collapse on this line is correct and well-targeted for the observed ci-scan bug. Two consistency gaps remain between the SR title-cell escaping here and the Preview engine's shared Format-MarkdownCell — both worth folding into the already-deferred “centralize SR cell escaping” follow-up:

1. Angle brackets (credit: Gemini). This cell escapes | and now collapses \r\n, but does not escape </>. Format-MarkdownCell does, and the author's own comment there states GitHub “would otherwise swallow List<T> as an HTML tag.” So a ci-scan title containing angle brackets renders with reduced fidelity here. (Framing it as content-fidelity/consistency rather than injection — GitHub sanitizes rendered HTML, so this isn't a security hole, and the PR is correctly scoped to the live newline bug.)

2. Sibling SR sites with no escaping at all. Three other sites embed upstream titles into pipe-delimited rows with neither pipe-escape nor newline collapse: L3196 $issCell = "$issLink — $titleShort" ($r.title), L3386 AppendLine("| $prLink | $title | … |") ($pr.title), L3439 AppendLine("| $issueLink | $title | … |") ($it.title). An unescaped | in a PR/issue title is legal and common and would already break those rows today.

Net: the PR is fine as-is for the observed bug. The “only cell fed unsanitized upstream titles” wording in the description slightly understates the existing surface, and a future Format-MarkdownCell-style SR helper should cover all of these sites and ideally match Preview's full pipe/newline/angle-bracket contract.

[kubaflo]

Note

🤖 This review was automatically generated by a multi-model AI review system (Claude Opus 4.8, GPT-5.5, Gemini 3.1 Pro). Three models independently reviewed the code, then cross-pollinated their findings to reconcile differences. This is Round 2 following significant scope expansion (+34/-2 → +261/-16) after Round 1 LGTM.

Multi-Model Review Summary — PR #36031 Round 2

Title: Sanitize upstream titles in release-readiness tracker tables (embedded newlines + unescaped pipes)
HEAD: 0b7f7555 · Stats: +261/-16 lines, 3 files

Verdict: 🟡 NEEDS_DISCUSSION

Confidence: Medium
Recommendation: Code is ship-quality (unanimous), but required CI check is skipping due to by-design path filter for .github/**-only changes.

---

Consensus Assessment (3/3 Models)

✅ Code Quality: Unanimous LGTM on Merits

All three models independently concluded the code has no defects:

• Sanitizer correctness verified — Format-MarkdownTableCell (CR/LF collapse + pipe escape) is correct and null-safe; replacement order prevents dangling \ on truncated strings
• Completeness confirmed — Every upstream-title cell across both tracker engines routes through sanitization; cells that bypass it carry only engine-generated data
• Security rationale validated — Deliberate <> omission in SR (preserving List<T> fidelity) is defended by (a) hash-freeze immunity via head -n1 extraction, (b) marker-forgery immunity via full-line-anchored regex + newline collapse
• Tests are discriminating — Opus empirically verified: neutering Format-MarkdownTableCell caused exactly 21 sanitization assertions to fail; "coarse sanity" checks correctly stayed green
• Round 1 suggestions addressed — Consolidation into shared helper complete; test clarity comments added

---

CI Status Disagreement

The one divergence: How to handle by-design CI skip for .github/**-only PRs.

Check	Status	Why
license/cla	✅ pass	—
maui-pr (required)	⚠️ skipping	eng/pipelines/ci.yml path filter excludes .github/** (line 30) — all 3 changed files are under .github/skills/

Opus investigated:

• Path-filter skip is deterministic and correct for tooling-only PRs
• No workflow runs Test-ReleaseReadiness.ps1 — skill-validation reports "passed (static only)"
• Functional validation is local-only (Opus ran 568 tests + confirmed discrimination)
• Position: LGTM / Medium confidence — code is sound, confidence capped because no automated functional gate

Gemini & GPT:

• CI skip means undetermined/no coverage
• Policy: Never LGTM with high confidence when CI is skipped/pending
• Position: NEEDS_DISCUSSION / Low-Medium confidence — wait for CI to run before LGTM

Synthesis: Per multi-model verdict resolution rules, use majority verdict when split. Majority (2/3) is NEEDS_DISCUSSION, so final verdict reflects CI policy conservatism. Note for merger: This is routine for .github/**-only PRs — maui-pr build has nothing to compile. The 29 new Pester tests pass locally (Opus-verified) but have no CI coverage.

---

Findings

💡 Suggestion (Non-Blocking) — Defense-in-Depth

File: .github/skills/release-readiness/Get-ReleaseReadiness.ps1
Lines: 3484–3488 (regression table), 3350 (CI-pipeline), 3380 (Reverts table)

$title cells correctly route through Format-MarkdownTableCell, but recommendedAction and some other cells are interpolated raw:

"| … | $($it.recommendedAction) |"  # L3488

Safe today — recommendedAction values (L1914–2309) are engine-generated constants/issue numbers only.

Future-risk — If a future change embeds an upstream title in recommendedAction or other currently-raw cells, it would silently bypass sanitization.

Optional mitigation: Route all table cells through Format-MarkdownTableCell, or add comments marking cells as "intentionally engine-controlled."

---

Round 2 Scope Expansion Assessment

Author expanded from targeted ci-scan fix (+34/-2) to comprehensive, render-layer sanitization (+261/-16) addressing both Round 1 suggestions:

1. SR escaping consolidation ✅ — All scattered -replace '\|' sites centralized into Format-MarkdownTableCell helper. Deliberate <> omission for SR (vs. Preview's escaping) is documented and defended.

2. Test clarity ✅ — Added explicit "Discrimination note" and relabeled assertions (coarse sanity; non-discriminating) vs. (discriminating regression guard).

Result: 29 new tests (539 → 568), single choke point for all upstream-title cells, comprehensive newline-collapse + pipe-escape contract.

---

Bottom Line

Code verdict: ✅ Ready to ship (3/3 unanimous)
CI caveat: Required check is skipping by design (.github/** path filter)
Action: Maintainer can merge with standard .github/**-only-PR workflow, noting that Pester tests have no CI coverage but were verified locally by review automation.

Confidence: Medium — full marks on code quality, CI skip is expected but lacks automated functional gate.

[kubaflo]

💡 💡 [Suggestion] Defense-in-depth: Engine-controlled cells bypass sanitizer

$title correctly routes through Format-MarkdownTableCell, but recommendedAction (L3488), CI-pipeline cells (L3350), and Reverts table cells (L3380) are interpolated raw.

Safe today — All contain only engine-generated constants/issue numbers (L1914–2309), never upstream titles.

Future risk — If someone later embeds an upstream title in these cells, it would silently bypass sanitization and re-open row-split/pipe-injection vectors.

Optional mitigation: Route all table cells through Format-MarkdownTableCell, or add comments marking them as "engine-controlled, intentionally unsanitized."

(Opus finding, validated by Opus cross-poll, not contradicted by peers)

[kubaflo]

Note

🤖 This review was automatically generated by a multi-model AI review system (Claude Opus 4.8, GPT-5.5, Gemini 3.1 Pro). Three models independently reviewed the code, then cross-pollinated their findings to reconcile differences. This is Round 3 reviewing incremental changes after Round 2 NEEDS_DISCUSSION verdict.

Multi-Model Review Summary — PR #36031 Round 3

Title: Sanitize upstream titles in release-readiness tracker tables (embedded newlines + unescaped pipes)
HEAD: 4f7d8796 (Round 3) · Stats: +299/-18 lines, 3 files

Verdict: ✅ LGTM

Confidence: High (Unanimous 3/3)
Recommendation: Ready to merge. All Round 2 findings addressed, CI validated.

---

Unanimous Consensus (3/3 Models After Cross-Pollination)

Round 3 represents excellent responsive development:

All three models independently reviewed the Round 3 delta (0b7f7555 → 4f7d8796, +61/-25 net across 2 commits) and reached unanimous LGTM with high confidence after cross-pollination.

What Changed in Round 3

1. Round 2 Defense-in-Depth Finding: ✅ ADDRESSED

Round 2 flagged that recommendedAction and other engine-controlled cells bypassed the sanitizer. Author responded by:

• Added <> escaping to Format-MarkdownTableCell — Now provides full SR↔Preview parity
• Side benefit: Fixes latent display bug for Bump <PatchVersion> milestone titles
• All upstream-controlled cells now route through sanitizer with complete escape coverage

2. New Proactive Security Fix: Escape-the-Escaper Protection

Author discovered and fixed a "pipe-breakout" vulnerability where a title like foo\|bar would:

• Attempt escape: foo\|bar → foo\\|bar (double backslash + escaped pipe)
• But \\ renders as literal \, leaving unescaped | → table column injection

Solution: Backslash-doubling regex (\\*)\| → $1$1\|

• Doubles all backslashes immediately preceding pipes
• Example: foo\|bar → foo\\\|bar (3 backslashes: \\ renders as \, then \| renders as |)
• Mathematically sound: N backslashes → 2N+1 (odd count) → pipe always inert
• Perfect fidelity: Non-pipe-adjacent backslashes unchanged

Refined across 2 commits after GPT-5.5 flagged initial over-aggressive fidelity impact.

Empirical Verification (Opus)

• Dot-sourced both sanitizers with 12 adversarial edge cases (nested escapes, Unicode, multi-space runs)
• Byte-identical output with pre-expansion behavior on non-adversarial inputs
• Full test suite: 566/0 pass under TZ=UTC (CI environment)
• T8 local failure: Pre-existing timezone-fragile test (not in Round 3 delta, green in CI)

CI Validation — Critical Clarification

Round 2 incorrectly stated: "No workflow runs Test-ReleaseReadiness.ps1"

Round 3 correction (Opus discovery):

• release-readiness.yml Validate (PR) job DOES run the full test suite
• Triggers on pull_request for .github/skills/release-readiness/** (exact match for this PR)
• Status: ✅ Passing (1m47s) on HEAD 4f7d8796
• Tests: 566/0 pass in CI's UTC environment

Why maui-pr skips:

• eng/pipelines/ci.yml path filter excludes .github/** (line 30)
• By design: maui-pr builds MAUI product code (Core/Controls/Essentials)
• These are PowerShell tooling scripts with nothing to compile
• Correct skip, proper validation via Validate job

GPT's verdict evolution:

• Independent: NEEDS_DISCUSSION (Low) — Cited gh pr checks --required showing maui-pr skip
• Cross-poll: Upgraded to LGTM (High) — After learning about Validate job passing

---

Findings

Zero blocking or warning-level findings from all three reviewers.

All three models returned empty inline findings arrays after cross-pollination. No regressions, no new issues, no concerns.

---

Round-by-Round Evolution

Round	Verdict	Key Outcome
1	Unanimous LGTM (High)	Targeted ci-scan fix, 2 non-blocking suggestions
2	Split → NEEDS_DISCUSSION (Majority)	Scope expansion addressed suggestions, unanimous on code quality, split on CI interpretation
3	Unanimous LGTM (High)	Defense-in-depth addressed, escape-the-escaper fix, CI validation confirmed

Total scope: Initial +34/-2 → Final +299/-18 (9× growth across 3 rounds)

---

Bottom Line

Code quality: Excellent responsive development. Author:

• Addressed all Round 2 feedback within ~1 hour
• Proactively discovered and fixed escape-the-escaper vulnerability
• Refined solution across 2 commits after review feedback
• Maintained perfect test coverage (29 new tests, all discriminating)

CI validation: Confirmed passing via release-readiness.yml Validate (PR) job

Security: Complete sanitization coverage (newline collapse + pipe escape + angle-bracket escape + backslash-doubling)

Verdict: Ship it. 🚀

Confidence: High — Unanimous agreement, empirically verified, CI validated, zero findings.