Docs to Corpus Pipeline - Nexus Knowledge Base

Nexus Docs (CopyParty) - File Upload & Pipeline

Overview

CopyParty provides the Nexus Docs web interface and file storage system. It serves as the entry point for user file uploads that can then be processed by AI and stored in Corpus.

1. File Upload Entry Points

Web UI

URL: docs.corlera.com
Authentication: Username/password
Features:
Drag-and-drop upload
Folder creation
File sharing with expiring links
Media previews (images, video, audio)

MCP Tools

docs.upload: Upload files from local paths or TEMP staging
docs.create_folder: Create directories
docs.move: Organize files between folders

2. Storage Locations

Visibility	Path	Access
Private	`/data/cdn/users/{user}/documents/`	User only
Public	`/data/cdn/users/{user}/public/`	Anyone with link
Shared	`/data/cdn/users/{user}/shared/`	Configurable

Base Path: /data/cdn/users/

3. AI Processing Pipeline

┌─────────────────────────────────────────────────────────────────┐
│                    DOCS → CORPUS PIPELINE                       │
├─────────────────────────────────────────────────────────────────┤
│                                                                 │
│  ┌──────────────┐    ┌──────────────┐    ┌──────────────┐      │
│  │   UPLOAD     │    │   PROCESS    │    │    STORE     │      │
│  │              │    │              │    │              │      │
│  │ docs.corlera │ →  │  LLMSherpa   │ →  │   Corpus     │      │
│  │   .com       │    │  (Port 6650) │    │   Server     │      │
│  │              │    │              │    │              │      │
│  │ /data/cdn/   │    │  Extract:    │    │  Quadfecta:  │      │
│  │ users/{user} │    │  - Text      │    │  - FTS       │      │
│  │              │    │  - Structure │    │  - Vector    │      │
│  │              │    │  - Tables    │    │  - Graph     │      │
│  │              │    │  - Images    │    │  - Relational│      │
│  └──────────────┘    └──────────────┘    └──────────────┘      │
│                                                                 │
└─────────────────────────────────────────────────────────────────┘

Pipeline Steps

Monitor: Watch /data/cdn/users/ for new uploads
Extract: Send to LLMSherpa (port 6650) for intelligent parsing
Process: LLMSherpa extracts text, tables, structure
Ingest: Store extracted content in Corpus with metadata
Index: Quadfecta indexes for multi-modal search

4. URL Mapping

Web URL	Local Path
`docs.corlera.com/home/chris/`	`/data/cdn/users/chris/`
`docs.corlera.com/home/chris/documents/`	`/data/cdn/users/chris/documents/`
`docs.corlera.com/home/chris/public/`	`/data/cdn/users/chris/public/`
`docs.corlera.com/shr/{share_id}`	`/data/cdn/shr/{share_id}`

5. Infrastructure

Component	Details
Container	`nexus-cdn`
Image	`copyparty/ac`
Port	`3923`
Config	`/data/cdn/system/config/copyparty.conf`
Domain	`docs.corlera.com`

6. Integration Points

TEMP Server: Upload from staging via temp_id parameter
Corpus Server: Destination for extracted content
LLMSherpa: Document extraction service (port 6650)
Search Server: Query indexed content via Quadfecta