Manage documents in a Paperless-ngx instance via its REST API — upload, search, read, and download documents.
| Skill | Description |
|---|---|
/paperless:configure |
Save the instance URL, username, and password; verify connection |
/paperless:upload |
Upload a local file to Paperless-ngx with optional metadata |
/paperless:search |
Full-text search, similarity search (more_like=<id>), or autocomplete |
/paperless:content |
Display the full OCR-extracted text of a document by ID |
/paperless:view |
Download the archived PDF for a document; when called from Telegram, sends the file directly to the chat |
The following keys are stored in ~/.claude/channels/paperless/.env:
| Key | Description |
|---|---|
PAPERLESS_URL |
Base URL of your instance (e.g. http://paperless-ngx:8000) |
PAPERLESS_USERNAME |
Your Paperless-ngx username |
PAPERLESS_PASSWORD |
Your Paperless-ngx password |
Run /paperless:configure to set them up interactively.
Note: The API user must own (or have been granted access to) the correspondents, tags, and document types you want resolved by name. If the
phoenixservice account cannot see metadata owned by another user, update the object-level permissions in Paperless-ngx or switch to the owner's credentials.
/paperless:search maintains a local cache of resolved IDs at
~/.claude/channels/paperless/id_cache.json to avoid redundant API calls.
The cache maps correspondent, document type, and tag IDs to their names.
Unresolvable IDs are stored as "?" so they are not re-fetched.
/plugin marketplace add ~/Workspace/paperless-skill
/plugin install paperless@paperless
/reload-plugins
MIT