Question: review queue summary for issue-focused PRs

[YOMXXX]

Question

GitHub Discussions appear to be disabled for this repository, so I am using the Question template as a short coordination summary for a batch of issue-focused PRs.

All PRs below are open and currently have passing GitHub checks. A few have duplicate-bot comments; I added notes on the PRs explaining why the implementation or test coverage is still distinct.

PR	Issue	Area	Status	Notes
#29297	#29291	shell	checks passing	Duplicate bot points to #29288. This PR adds stronger regression assertions that metadata preview actually truncates by UTF-8 byte length and does not start on an unpaired low surrogate.
#29296	#29292	config	checks passing	Duplicate bot points to #29289. This PR also covers ConfigManaged.readManagedPreferences(), not only the final default username assignment.
#29295	#29220	provider/MCP	checks passing	Inlines local $ref schemas for OpenAI-compatible custom providers and handles recursive refs for MCP tool schemas.
#29293	#29262	TUI	checks passing	Avoids the placeholder dummy session route during --continue --fork.
#29239	#29054	task tool	checks passing	Duplicate bot points to #29048. This PR includes regression coverage for empty subagent text output failing instead of returning an empty successful task result.
#29238	#29200	HTTP/config errors	checks passing	Duplicate bot points to #29208. This PR keeps config failures intact and maps them at the HTTP boundary to structured 400 responses.
#29236	#29203	VCS diff	checks passing	Stabilizes git diff patch prefixes.
#29230	#29214	shell tests	checks passing	Waits for shell output drain to reduce terminal fallback flake.
#29225	#29041	config permissions	checks passing	Allows web permission object rules.

Is this PR grouping and duplicate-handling approach useful for maintainers, or would you prefer closing overlapping PRs in favor of earlier simpler ones even when the newer PR has narrower coverage or stronger tests?