[compass] Anode 0.11.2 release candidate, so a fixed build follows the launch within days

[skulitom]

What

The 0.11.2 release candidate: the version change, changelog and release notes for everything merged since 0.11.1, a 0.11.2 section in docs/RELEASE-READINESS.md with the owner's live check as exact commands, and a live check script for #28 and #29. It follows CONTRIBUTING's release process, steps 1 to 4. Merging it publishes nothing; the tag and the release follow the owner's live check.

Why

main has 14 bug fixes that 0.11.1 users don't have (#10 to #29: typing that garbled long text in Chrome, set_value reporting success on a <select>, seat_observe failing on File Explorer and repeating Chrome's tree, lost replies behind long operations, and more), plus the faster startup. A fix release soon after the launch is cheap trust. The live check needs a real seat, so it's the owner's; this PR makes it one sitting of about 25 minutes.

Changes

• Version 0.11.2 in src/Anode/Anode.csproj, server.json, packaging/mcpb/manifest.json and .claude-plugin/plugin.json. CHANGELOG's Unreleased section becomes ## 0.11.2 — 2026-10-09: release.yml runs test-distribution.ps1 -Release, which needs a dated ## 0.11.2 heading and no unassigned Unreleased changes.
• .github/RELEASE_NOTES.md, the release body: what's new, the known issues with their workarounds, validation, and the upgrade steps.
• docs/RELEASE-READINESS.md: a new 0.11.2 section at the top. What was checked without a seat; the owner's live check (build the candidate, install it, run test-desktop.ps1, test-chrome-form.ps1, test-development.ps1 -InstallBrowserTools -VerifyInput and a File Explorer inspect, then tag); what stays covered only by stand-ins; and the known open bugs. The candidate is found with git log -S '<Version>0.11.2</Version>', so fixes merged after this PR wait for the next release. The 0.11.1 record below it keeps its text; its headings move down a level.
• scripts/test-chrome-form.ps1 (new, opt-in, like test-desktop.ps1): in a Chrome --app window on a local page with a temporary profile, it checks that set_value on a text field still succeeds and reads back, that set_value on a <select> either changes the choice or fails with [compass] Fix: seat_element set_value reports "performed" on a Chrome drop-down list that ignores it #28's "still showed its old value" message, never a false success, and that an observation with the list open lists the text field once ([compass] Fix: seat_observe repeats a window's tree while a Chrome drop-down list is open #29). It also reports whether the open list's options appear in the tree (a warning, not a failure), and chooses one when they do. It closes only its own Chrome window and deletes only its own temporary profile. docs/DEVELOPMENT-TESTING.md lists it with the other live checks.
• Not changed yet, on purpose: bucket/anode.json and packaging/winget/* (they need the release's SHA-256, step 7), docs/INSTALL.md's pinned -Version example and the README's "Known issues in 0.11.1". Changing them now would point main at a release that doesn't exist yet; they follow the release in one PR.

Publishing safety

• The MCP Registry lists Anode 0.11.0 (metadata only), read today. On this push to main, publish-registry.yml sees the listing and does nothing. Its daily run uses the latest release, v0.11.1, which has no .mcpb asset, so it does nothing either. After the 0.11.2 release it publishes the installable entry only once the tested bundle is attached.
• release.yml runs only on a v* tag.

How it was tested

On this commit's tree, in a separate output folder:

• scripts\build.ps1 -OutputDirectory artifacts\pkg-build -ArchiveDirectory artifacts\pkg-release -QuickTest -Package -Mcpb: 97/97 quick checks; documentation consistent (36 documents, 41 tools).
• scripts\registry-entry.ps1 -Tag v0.11.2 on the built bundle, then scripts\test-distribution.ps1 -ArchiveDirectory artifacts\pkg-release -Release: all distribution checks pass, including the CHANGELOG and release-notes checks the Release workflow runs.
• test-chrome-form.ps1 parses under Windows PowerShell 5.1. It has not run: it needs a live seat, and no seat, lease, browser or Android command was run here. Its first run is the owner's live check.
• The .NET runtime pin, 10.0.12, is still the latest 10.0 patch (Microsoft's release metadata, read today).
• Rebuilt after the review fixes: the package's docs match the commit, and the checks above pass again on it.
• Second review: a fresh read-only Claude reviewer (Codex is at its usage limit until 22:15 today) checked every claim against the code, the workflows and every command and script parameter in the owner's steps. It found one P1: test-chrome-form.ps1 would always fail under Windows PowerShell 5.1, where a single PSCustomObject has no Count. Fixed in e89e79d (results wrapped in @(...), confirmed in PS 5.1.26100 with a mock observation), with its P2 (the release notes now link the validation record on main, where the results go) and P3s (scope wording, launch inside try with cleanup of every Chrome process the script started, an Explorer null check, and the README's 0.11.1 known issues reworded so they stay true inside the 0.11.2 package). Its re-check of e89e79d: "no credible problem".

🤖 Generated with Claude Code

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-09T08:48:42.762837Z	e89e79d	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: e89e79d7c2

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Update the bundled installer example for 0.11.2

When this version bump is tagged, scripts/build.ps1 copies the docs directory into the 0.11.2 archive, but docs/INSTALL.md:40 still tells users under “Choose a version” to run install.ps1 -Version 0.11.1. Anyone following the new package's explicit-version example will therefore install or downgrade to 0.11.1; changing the documentation after the release cannot repair the tagged source or published archive, so the pinned example needs to be updated in this candidate.

Useful? React with 👍 / 👎.