Paperless

**Judgement:** Strong, focused integration for environments that already use Paperless-NGX; highly suitable for building document-centric assistants, provided the runtime has `ppls` installed and configured. **What it does well:** - Exposes most common Paperless-NGX operations (search, list, show, download, upload, update metadata, manage tags/correspondents/doc types). - Designed for automation via `--json`, which makes parsing results straightforward for an agent. - Filter model is expressive (tags, date ranges, correspondents, document types, missing metadata, recent changes), enabling nuanced retrieval workflows. **Key risks / constraints:** - Requires a correctly configured `ppls` CLI (hostname + API token) and network access to the Paperless-NGX instance; misconfiguration will cause tool failures. - Operates on potentially sensitive documents (invoices, contracts, receipts); improper prompt design could lead to over-broad searches or unwanted disclosure/download of documents. - Write operations (uploads, metadata updates, creating tags/correspondents/types) can unintentionally clutter or misorganize the archive if the agent’s logic is faulty. - No explicit rate/size controls are visible; bulk downloads or large queries may stress the Paperless-NGX server if not guarded. **Recommended scenarios:** - Self-hosted, trusted environments where the user explicitly wants an assistant to search and manage their Paperless-NGX archive. - Workflows that primarily need read-only search and document retrieval, with carefully gated use of upload/metadata mutation. - Technical users or teams comfortable managing API tokens and CLI tools, who want a programmable layer over their document store for higher-level automation or conversational access.