[mblua]

Closes #1606.

Documentation catch-up for two changes that landed today. No behavior, workflow, or script change.

What was wrong

The portable download had no name. #1589 shipped agentscommander-X.Y.Z-windows-x86_64-portable.zip, gated so a release without it stays a draft. But docs/quickstart.md still offered "the portable agentscommander.exe", which has never been a published asset. A reader following that instruction downloads the raw agentscommander-windows-x86_64.exe instead, and that name parses as instance suffix 64, silently giving them .agentscommander-windows-x86_64\, an AC [64] badge, a non-default mutex, and ports 9886/9906 instead of 9877/9887. Naming the real asset is what prevents that.

The second-instance behavior changed. #1592 replaced the silent exit(0) with a message explaining the collision and pointing at the rename. packaging/windows/PORTABLE.txt ships inside the zip specifically to explain that scenario and still described the old behavior.

Deliberately not fixed here

Two claims in these same files are inaccurate today. Both are fixed by work in flight, so correcting them now would mean writing the text twice:

Claim	Why it waits
README.md:63 "copy the .exe to any drive and it carries its own config"	True only where the destination is writable. #1577 adds the probe and the fallback.
docs/features/portable-instances.md:85 "AC keeps no global state outside that directory"	The WebView2 user-data folder defaults to %LOCALAPPDATA%\dev.agentscommander.app\ because no builder sets data_directory. #1590 confines it.

macOS and Linux portability wording is untouched for the same reason: each needs #1577 plus a platform-specific bundle-aware resolver branch (#1594 for macOS; the AppImage case is not filed yet). Until those land, an AppImage or a .app is a per-user install rather than a portable one, and the docs should not claim otherwise in either direction.

Verification

• Four files changed, 10 insertions, 7 deletions.
• The two deferred claims are confirmed present and unmodified.
• No page anywhere in README.md or docs/ still offers a portable download by a name that is not a published release asset.
• bundle-validation runs on this PR because it touches packaging/windows/**, so the reworded PORTABLE.txt is re-packed and re-smoked against a real built binary.

https://claude.ai/code/session_01HydthWuftnNBoYzqHZTNm8

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[mblua]

Heads-up: the required checks on main changed (#3184, PR #3185).

The main ruleset now requires these checks instead of the old names:

• rust-regression is now rust-regression-windows
• windows-release-cli-smoke is now release-cli-smoke-windows
• new: issue-1850-release-windows, issue-1850-release-linux, issue-1850-release-macos and rust-slow-test-windows

This branch still runs the old workflow, so those checks never report here and the PR cannot merge as it is.

What to do when this PR is ready to land: wait until PR #3185 is merged, then bring this branch up to date with main (the ruleset already requires that). The new workflow comes with it and reports the new check names. Resolve any conflict in .github/workflows/pr-regression-gates.yml in favour of main.