What problem does it solve? Research platforms accumulate orphaned uploads — duplicate datasets, abandoned pending uploads, files uploaded by mistake — that clutter storage and search results. This Skill permanently removes those unlinked Chronicle assets while protecting anything still referenced by an experiment record. ## Core Features & Use Cases - Guarded hard delete: Calls the bundled chronicle.delete_asset MCP tool, which transactionally removes the asset row, ACLs, and inline content, then purges storage bytes and search documents best-effort. - Linked-asset protection: The server refuses deletion with a 409 (including per-table link counts) when an asset is referenced by any experiment, variation, or output — those assets are governed by deprecate/invalidate instead. - Confirmation-first workflow: Always previews the exact asset list and requires explicit user confirmation before the irreversible delete. - Use Case: After deleting a draft experiment, its formerly linked datasets become orphans; use this Skill to clean them up, sorting outcomes into deleted, linked-skipped, already-gone, and failed. ## Quick Start Ask the agent to permanently delete the orphaned Chronicle assets with these ids, confirming the previewed list before the irreversible deletion proceeds.
