Document Workflow Implementation Summary

Date: December 17, 2024 Track Project: df8448e7 (50% complete)

What Was Accomplished

1. Environment Renaming (Memory → Corpus)

Renamed /opt/mcp-servers/memory/ → /opt/mcp-servers/corpus/
Renamed mcp_memory_server.py → mcp_corpus_server.py
Changed class MemoryMCPServer → CorpusMCPServer
Changed prefix memory: → corp:
Changed stable ID prefix m_ → c_
Migrated 107 Redis keys from docs: to corp:
Updated 103 documents with new c_ stable IDs

2. Moved create_pdf to Nexus Docs

Added create_pdf tool to /opt/mcp-servers/docs/mcp_docs_server.py
Added TEMP integration (temp_id parameter)
Supports pulling content from TEMP staging
Auto-cleanup of TEMP after PDF creation
Removed create_pdf from Corpus MCP (was duplicate)

3. Document Workflow Architecture

Environment	Port	Prefix	Stable ID	Purpose
TEMP	6680/6681	temp:	tmp_XXXX	Staging, 24h TTL
Corpus	6650/6651	corp:	c_XXXX	Searchable text, Quadfecta
Nexus Docs	CopyParty 3923	(paths)	(files)	Files, PDFs, sharing

4. KB Documentation Created

fe0a50bc: Document Workflow Architecture (main reference)
Updated 998d6211: Corpus Environment
Updated e732b2cc: Nexus Docs Environment

Key Files Modified

/opt/mcp-servers/corpus/mcp_corpus_server.py - v2.1.0
/opt/mcp-servers/docs/mcp_docs_server.py - v2.1.0

Workflow: Research → Document → PDF

1. Research → AI gathers from web, corpus, kb, contacts
2. Stage → temp.stage(content) returns tmp_XXXX
3. Commit → docs.create_pdf(temp_id=tmp_XXXX) creates PDF
4. Index → corpus.create() stores searchable text (optional)
5. Cleanup → TEMP auto-cleaned after PDF creation

Why This Architecture

Speed: TEMP staging avoids streaming large content through chat
Separation: Corpus = AI knowledge, Docs = Human files
Searchability: Quadfecta indexing on Corpus entries
Multi-tenant: User parameter required, no hardcoded defaults
Future: 'Memory' name reserved for true RAG/temporal/graph system

Remaining Work

CopyParty password-protected sharing (Track: df8448e7)