Release 1.1.2 — the default path works end to end

Fix release for 1.1.1: the minimal path end to end, engine commands that
run on Godot, Unity and Unreal, hook warnings you can see, and verdicts
that say when they could not check. Details in CHANGELOG.md; upgrade
steps in UPGRADING.md (v1.1.1 → v1.1.2).

Fixes #67
Fixes #131

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01WuJn2FdynEybfL5KC697nJ

.claude/hooks/yaml-helper.sh

@@ -673,7 +673,8 @@ EOF
 #   checkpoint that `.claude/docs/context-management.md` tells users to rely on.
 #   Shipping `off` as the default would have silently removed crash recovery,
 #   the session archive and the subagent spawn tally from every existing
-#   project. Users who want the ~2-5k tokens per session opt out explicitly.
+#   project. Users who want the checkpoint preview's tokens back (at most ~750
+#   per session start) opt out explicitly.
 #
 #   Deliberately awk, not get_effective_yaml_key: this runs in log-agent.sh on
 #   EVERY subagent spawn, and get_effective_yaml_key shells out to Python twice.
@@ -1240,9 +1241,14 @@ resolve_code_root() {
   esac
 
   # 3. No usable engine value — believe the tree, but only if it is unambiguous.
+  #    Names are compared exactly, against the directory listing: `[ -d Assets ]`
+  #    is also true for Godot's own `assets/` on a case-insensitive filesystem
+  #    (Windows, default macOS), which made src/ + assets/ "ambiguous" and a lone
+  #    assets/ a Unity project.
   if [ -z "$root" ]; then
-    for d in src Assets Source; do
-      if [ -d "$_YH_ROOT/$d" ]; then found="$d"; n=$((n + 1)); fi
+    for d in "$_YH_ROOT"/*/; do
+      d="${d%/}"; d="${d##*/}"
+      case "$d" in src|Assets|Source) found="$d"; n=$((n + 1)) ;; esac
     done
     if [ "$n" = 1 ]; then root="$found"; src="detected"; else root=""; src=""; fi
   fi
@@ -1464,7 +1470,8 @@ resolve_config() {
   #      "use as-is", while the notes line simultaneously claimed it had been
   #      "ignored, chain continued". resolve_setting falls invalid values
   #      through to the default, which is what every other knob already did --
-  #      `review_mode: nonsense` resolves to `lean (rigor:standard)`.
+  #      `review_mode: nonsense` resolves to the rigor value, `solo (rigor:minimal)`
+  #      by default.
   #   2. No provenance. Every other line names its source; this one printed a
   #      bare value with a `${pe:+}` suffix that expands to nothing -- a
   #      provenance tag someone started and never finished. The key is on the
@@ -1546,6 +1553,28 @@ EOF
   return 0
 }
 
+# --- hook_warn: an advisory warning that someone actually sees ---------------
+#
+#   hook_warn <PreToolUse|PostToolUse> "<message>"
+#
+# A hook that writes a warning to stderr and exits 0 is writing to the debug
+# log: Claude Code shows that text to neither the user nor Claude. Hook JSON on
+# stdout reaches both -- systemMessage is shown to the user, additionalContext
+# is given to Claude next to the tool result. It sets no permissionDecision:
+# "allow" would skip the user's permission prompt. stdout must hold ONLY this
+# object, so call it once, as the hook's last output, then exit 0.
+#
+# There is no jq to lean on, so the string is escaped here: backslash and quote
+# by sed, tab and newline by awk. CR and other control characters are dropped.
+hook_warn() {
+  local m
+  m=$(printf '%s' "$2" | tr -d '\000-\010\013-\037' \
+      | sed 's/\\/\\\\/g; s/"/\\"/g' \
+      | awk '{ gsub(/\t/, "\\t"); if (NR > 1) printf "\\n"; printf "%s", $0 }')
+  printf '{"systemMessage":"%s","hookSpecificOutput":{"hookEventName":"%s","additionalContext":"%s"}}\n' \
+    "$m" "$1" "$m"
+}
+
 # --- Direct execution: the skill-bootstrap entry point -----------------------
 #
 #   bash "${CLAUDE_SKILL_DIR}/../../hooks/yaml-helper.sh" resolve_config --keys a,b
@@ -1559,7 +1588,7 @@ EOF
 # `${CLAUDE_SKILL_DIR}` is substituted as text before the check, so one plain
 # `bash <path> resolve_config …` call is approvable by the matching grant in the
 # skill's own `allowed-tools`. That is also what lets a Bash-less agent preload
-# the skill (GitHub issue #128).
+# the skill.
 #
 # Only resolve_config is dispatchable: the bootstrap is the one caller, and
 # every name added here widens what a skill grant can reach.