fix(docs-sync): use awk to extract only first frontmatter block (#19)

The sed range pattern '/^---$/,/^---$/p' restarts after termination,
capturing every '^---$' pair in the file. Files with markdown HR
separators in the body (api-reference.md upstream has 65) had their
content doubled on each sync, then re-doubled on the next.

Replaced with an awk single-state-machine: increments n on each ---,
prints the line, exits at the second ---; in between, prints body
lines. Idempotent for files with body HR separators; behavior
unchanged for files without (mcp.md, platform.md, etc.).

Verified by re-running the sync after the fix:
  Sync complete: 0 updated, 4 unchanged, 0 failed
api-reference.md stays at 1861 lines (was doubling to 2906 with the
sed pattern). Closes #19.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Author: Kurt Overmier

scripts/docs-sync.sh

@@ -237,8 +237,12 @@ ${generated}"
   if [[ "$has_frontmatter" == "false" ]]; then
     existing="${CONTENT_DIR}/${page_key}"
     if [[ -f "$existing" ]]; then
-      # Extract existing frontmatter and prepend to new content
-      existing_fm=$(sed -n '/^---$/,/^---$/p' "$existing")
+      # Extract existing frontmatter (first --- block only) and prepend.
+      # awk single-state-machine: include lines starting with the first
+      # --- and stop after the second one. Robust against bodies that
+      # use ^---$ as markdown horizontal-rule separators (would otherwise
+      # double on each sync — see docs#19).
+      existing_fm=$(awk '/^---$/{n++; print; if(n==2)exit} n==1 && !/^---$/{print}' "$existing")
       content="${existing_fm}
 
 ${content}"