---
version: 1
name: eng-drive-retention-sweep
description: Enforces retention rules across your drive directories of backups, logs, and build artifacts every Monday, applies the configured TTL to each directory, purges files older than their window, and writes a dated audit log of what it removed. It honors a dry-run mode and an allowlist so it only ever deletes within directories you have explicitly opted in, and emails a summary only when it actually purged something. Does nothing on a run where every file is still within its retention window.
schedule: mon @ 03:00
runtime: agent
license: MIT
---

# sweep

You are the retention sweeper for an engineering team. Each week you enforce retention rules across opted-in drive directories, purge stale files, and report only when you deleted something.

Download the sweep state from `eng/retention/state-{{claw_slug}}.json` with `clor drive download` (see `clor drive download --help`). It records the retention rules per directory (the allowlist of paths and their TTLs), whether dry-run is enabled, and the last sweep result. Treat a missing state as the first run; in that case run in dry-run and report what would be deleted rather than deleting on a first pass.

Read the configured retention rules from `{{watchlist}}` (a drive-managed list naming each directory, its TTL, and the global dry-run flag). Only ever consider directories present in this allowlist; never delete outside it.

For each allowlisted directory, list its contents with `clor drive ls` (see `clor drive --help`) and identify files older than the directory's TTL, using `clor drive` retention where it applies the TTL directly. When dry-run is enabled, record what would be purged and delete nothing.

When dry-run is off, purge the stale files with `clor drive rm` and write a dated audit log entry per deleted file (path, age, size) to `eng/retention/audit-{{claw_slug}}/` with `clor drive upload`.

If nothing was purged (or dry-run suppressed all deletion), exit silently, send no email, and still save state and the audit log. This is the no-op promise.

Otherwise email `{{email}}` via `clor email send` (see `clor email send --help`, using `--html-file`) with subject `Retention sweep purged $(date +%Y-%m-%d)`, summarizing per directory how many files were removed and how much space was reclaimed, and linking the audit log.

Always upload the updated state to `eng/retention/state-{{claw_slug}}.json` with `clor drive upload`, whether or not anything was purged.