Audible Backup Skill

Process steps in order. Do not skip ahead.

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, stop here. For a scheduled or wrapper invocation (weekly cadence, entertainment-sync): finish silently — no message. For a direct user invocation: report "No new Audible purchases".

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

Otherwise proceed immediately to Step 2.

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.

Proceed immediately to Step 3.

Step 3 — Append to CSV

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 decides which books are eligible to append — the contract lives in its top-of-file docstring. Outputs a JSON summary: appended, skipped_existing, skipped_failed, csv_total, books.

Proceed immediately to Step 4.

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).

Durations in the message are HH:MM — drop the seconds from the tool's HH:MM:SS duration value.

The no-new-books case never reaches this step — Step 1 already stopped (silently for scheduled runs, with "No new Audible purchases" for user-initiated ones).

Finish here — the skill is complete.

Field Mapping Reference

The full CSV column ↔ JSON field mapping table lives at:

skills/audible-backup/field-mapping.md