Every step below is mandatory. Execute them in order. Do not skip, reorder, or abbreviate any step.

Step 1: Run audible backup dry-run

Call the mcp__nanoclaw__audible_backup MCP tool with dryRun: true.

Parse the response. If new_books is 0 or books array is empty, report "No new Audible purchases" and stop.

If the tool errors (auth failure, Docker issue), report the error and stop.

Step 2: Download new books

If Step 1 found new books, call mcp__nanoclaw__audible_backup with dryRun: false to download and decrypt them.

Parse the response. Note which books succeeded (status: "ok") and which failed.

Step 3: Append to CSV

For each book with status: "ok", pipe the full backup response JSON to the CSV append script:

echo '<JSON>' | python3 /home/node/.claude/skills/tessl__audible-backup/scripts/csv-append.py

The script deduplicates by ASIN. Outputs JSON summary with appended count.

Step 4: Report results

Send a message summarizing what was downloaded. Format:

<b>Audible Backup</b>

• <b>Title</b> by Author (Series Name #N) — HH:MM
• <b>Title</b> by Author — HH:MM

N new books added to library (total: M).

If nothing new → silence (for scheduled runs). If user-initiated → "No new purchases."

Field Mapping Reference