v8.11.0: JSON and CSV Document Loaders
This release implements the missing JSON and CSV document loaders that were previously advertised but not functional, resolving Issue #181.
๐ฏ Key Features
JSON Document Loader
โ Nested Structure Flattening - Converts complex JSON to searchable text
- Configurable strategies: dot notation (
config.db.host) or bracket notation (config[db][host]) - Customizable depth limits to prevent deep recursion
- Optional type information inclusion
โ Flexible Array Handling
- Expand mode: Full array item expansion with indices
- Summarize mode: Array length summary for large datasets
- Flatten mode: Indexed array flattening
โ Use Cases
- Knowledge base exports (Roam Research, Notion, Obsidian)
- API response documentation
- Configuration file documentation
- Structured metadata archives
CSV Document Loader
โ Smart Auto-Detection
- Automatic delimiter detection (comma, semicolon, tab, pipe)
- Header detection with heuristic analysis
- Multi-encoding support (UTF-8/16/32, Latin-1, CP1252)
โ Efficient Processing
- Row-based text formatting with column context
- Configurable row chunking for large files
- Optional row numbering for traceability
โ Use Cases
- Data dictionaries and glossaries
- Reference tables
- Tabular documentation
- Log analysis results
๐ Bug Fixes
False Advertising Resolved - JSON and CSV were listed in SUPPORTED_FORMATS but had no implementations
- Before: Upload failed with "No loader available for file: document.json"
- After: Full functional support with proper chunking and metadata preservation
๐งช Testing
- 29 comprehensive unit tests added
- 15 tests for JSON loader (nesting, arrays, encoding, strategies)
- 14 tests for CSV loader (delimiters, headers, encoding, large files)
- All tests pass with full coverage of edge cases
๐ Integration
- Auto-registers with ingestion system on import
- Follows
DocumentLoaderbase class interface - Compatible with existing chunking and metadata systems
- Works with both HTTP API and MCP server document ingestion
๐ Related
- Closes: Issue #181 (JSON and CSV Support Advertised But Not Implemented)
- PR: #187 (Fix: Implement JSON and CSV document loaders)
- Merged: Oct 28, 2025
๐ Documentation
See CHANGELOG.md for complete release history.
Installation:
pip install --upgrade mcp-memory-service
# or
uv pip install --upgrade mcp-memory-serviceUsage Examples:
# Ingest JSON file
claude /memory-ingest config.json --tags documentation
# Ingest CSV file
claude /memory-ingest data.csv --tags reference
# Ingest directory with JSON/CSV files
claude /memory-ingest-dir ./data --tags knowledge-base๐ค Generated with Claude Code