exports-ops

name	exports-ops
description	Umbrella skill for private export and import pipelines. Use this when the task spans multiple data sources or when no more specific export skill applies.
memory_tags	["domain:exports","workflow:data-ingestion","skill_role:operator","repo_boundary:tools","data_class:private-derived","risk:high"]

Exports Ops

Trigger

Use this skill for reusable export/import automation:

• email/calendar exports
• Apple Notes and Cursor chat exports
• social feed ingestion/export
• private local screen-activity capture and derived contact sheets
• multi-source export workflows that touch more than one data source

Prefer the narrower skills when the intent is clear:

• x-check-ops
• email-ops
• calendar-ops
• apple-notes-export-ops
• cursor-chat-export-ops

Boundaries

• Specification source: liferepo workflow docs
• Private state source: <private-repo>/captures/ and <private-repo>/scripts/exports/* config/token files

Current Script Surface

Implementations currently live in:

• skills/exports-ops/scripts/

Large Google Drive archive transfers use scripts/download_google_drive_archive.py. It authenticates with Drive read-only scope, supports interrupted-transfer resume, verifies expected byte counts, and records optional SHA-256 checksums in an atomic manifest. Keep credentials, OAuth tokens, manifests, and downloaded files in the configured private state or external archive, never in this reusable skill repo.

Before transferring large ZIPs, use scripts/inventory_google_drive_zips.py to read their central directories via HTTP Range requests. This supports Drive-as-cold-storage workflows where only selected members are fetched later.

Use scripts/read_google_drive_zip_member.py with the generated inventory to preview or extract one exact member directly from Drive. Its stdout mode is bounded by default so an accidental read cannot flood the terminal or disk.

Use scripts/index_google_takeout_activity.py on a verified selected ZIP to turn My Activity HTML into a SQLite table plus FTS5 search index. Keep that derived private index beside the external archive rather than in this repo.

Use scripts/query_google_takeout_activity.py for bounded, read-only JSON searches against that index. Prefer its default result limit and add a service filter when the question is narrow.

Use scripts/index_meta_messages.py to build one atomic SQLite/FTS5 index directly from verified Facebook and Instagram JSON export ZIPs. Pass every ZIP part from one snapshot as --archive PLATFORM=ZIP; keep the output beside the private external archive. The indexer deduplicates repeated members, repairs common Meta UTF-8 mojibake, and does not extract the raw ZIP trees.

Use scripts/query_meta_messages.py for bounded, read-only full-text, platform, sender, thread, and UTC date-range queries. Prefer statistics or narrow filters before returning private message snippets.

Use scripts/index_imessage_messages.py on a consistent, read-only macOS Messages chat.db snapshot. The indexer uses the bundled Swift helper to decode legacy NSAttributedString message bodies, preserves attachment metadata without copying binaries into the derived index, and writes an atomic SQLite/FTS5 index. Keep both snapshot and index in private external storage.

Use scripts/query_imessage_messages.py for bounded statistics, timeline, chat, direction, service, and full-text queries. Use --content-only when a claim depends on words appearing in message text. Participant previews are capped so a group-chat query does not spill a full private roster.

Use scripts/export_imessage_daily_context.py for the separate daily-summary context lane. It must use SQLite backup semantics to create a consistent temporary copy of the live Messages database, query only that temporary copy, omit attachment bodies, filter automated verification traffic by default, and delete the database snapshot after extraction. The output is a short-lived, ignored private staging file; agents should write derived day-level context to the journal rather than raw handles or surprise message excerpts. This daily lane does not replace or advance the verified monthly archive checkpoint.

Use scripts/audit_export_freshness.py with a private freshness registry to check mixed automatic and snapshot exports without mutating their checkpoints. It understands per-account timestamp marker files and dated snapshot due dates; keep private account names and archive paths in the registry, not this repo.

Use scripts/capture_screen_activity.py for a pauseable macOS screen capture pilot. Raw images must go to a caller-configured private external archive; the collector must wait when that volume is missing rather than fall back to an internal disk. Keep the locked, idle, and sensitive-app exclusions enabled, use rolling raw retention, and treat generated contact sheets and image-token reports as analysis aids rather than permission to upload any screen content. The script makes no model or network calls automatically.

Legacy entrypoints remain in <private-repo>/scripts/exports/ as wrappers that delegate to this skill.