# Features Reference
This document provides an overview of all internal features and optimizations in the Obsidian MCP Server.
## Architecture Overview
```mermaid
flowchart TB
subgraph tools [MCP Tools Layer]
T1[search_vault]
T2[read_note]
T3[list_notes]
T4[Other Tools]
end
subgraph caching [Caching Layer]
FC[FileListCache]
CC[ContentCache LRU]
SC[SearchResultCache]
PC[PersistentCache LMDB]
end
subgraph optimization [Search Optimization]
II[InvertedIndex]
PT[PathTrie]
FP[FrontmatterParser]
SA[SearchScorer]
end
subgraph fs [Filesystem Layer]
V[(Obsidian Vault)]
FW[FileWatcher]
end
T1 --> SC
T1 --> II
T1 --> PT
T2 --> CC
T3 --> FC
CC --> PC
FC --> PC
SC --> PC
FW --> FC
FW --> CC
FW --> II
FW --> PT
V --> FW
```
## Feature Categories
### Caching Features
High-performance caching system with multiple layers.
| Feature | Purpose | Default |
| --------------------------------------------- | --------------------------- | --------------- |
| [File List Cache](file-list-cache.md) | Cache folder listings | 60s TTL |
| [Content Cache](content-cache.md) | LRU cache for file contents | 100 entries |
| [Search Result Cache](search-result-cache.md) | Cache search query results | 50 entries, 30s |
| [Persistent Cache](persistent-cache.md) | LMDB-backed persistence | Disabled |
| [Cache Warmup](cache-warmup.md) | Pre-load cache on startup | Disabled |
### Search Optimization Features
Advanced search capabilities with performance optimizations.
| Feature | Purpose | Default |
| --------------------------------------------- | --------------------------------- | -------- |
| [Inverted Index](inverted-index.md) | Fast word-to-files lookup | Disabled |
| [Path Trie](path-trie.md) | Efficient glob pattern matching | Disabled |
| [Frontmatter Parsing](frontmatter-parsing.md) | Tag and metadata extraction | Enabled |
| [Scoring Algorithm](scoring-algorithm.md) | TF-IDF-inspired relevance ranking | Enabled |
| [Search Compression](search-compression.md) | Compress large search results | Disabled |
### Monitoring Features
Real-time monitoring and cache invalidation.
| Feature | Purpose | Default |
| ------------------------------- | ------------------------- | ------- |
| [File Watcher](file-watcher.md) | Monitor vault for changes | Enabled |
## Data Flow
### Read Operation
```mermaid
sequenceDiagram
participant C as Client
participant H as Handler
participant CC as ContentCache
participant PC as PersistentCache
participant FS as Filesystem
C->>H: read_note(path)
H->>CC: get(path)
alt In memory cache
CC-->>H: content
else In persistent cache
CC->>PC: get(path)
PC-->>CC: content
CC-->>H: content
else Cache miss
CC->>FS: readFile(path)
FS-->>CC: content
CC->>PC: set(path, content)
CC-->>H: content
end
H-->>C: content
```
### Search Operation
```mermaid
sequenceDiagram
participant C as Client
participant H as Handler
participant SC as SearchCache
participant II as InvertedIndex
participant PT as PathTrie
participant FP as FrontmatterParser
participant FS as Filesystem
C->>H: search_vault(query)
H->>SC: get(queryKey)
alt Cache hit
SC-->>H: results
else Cache miss
alt Index enabled
H->>II: search(words)
II-->>H: candidate files
else
H->>FS: listFiles()
FS-->>H: all files
end
H->>PT: filterByGlob(files)
PT-->>H: filtered files
H->>FP: filterByTags(files)
FP-->>H: filtered files
H->>FS: searchContent(files)
FS-->>H: results
H->>SC: set(queryKey, results)
end
H-->>C: results
```
## Configuration Quick Reference
All features are configurable via environment variables. See [Configuration Reference](../configuration.md) for details.
```bash
# Enable all performance features
export OBSIDIAN_ENABLE_PERSISTENT_CACHE=true
export OBSIDIAN_ENABLE_INVERTED_INDEX=true
export OBSIDIAN_ENABLE_PATH_TRIE=true
export OBSIDIAN_ENABLE_CACHE_WARMUP=true
export OBSIDIAN_COMPRESS_LARGE_RESULTS=true
```
## Memory Usage Estimates
| Feature | Approximate Memory |
| --------------- | ---------------------------------- |
| Content Cache | ~10KB per cached file |
| File List Cache | ~100 bytes per file |
| Search Cache | ~1KB per cached query |
| Inverted Index | ~10-20 bytes per word-file mapping |
| Path Trie | ~200-500 bytes per file path |
## Performance Characteristics
| Scenario | Without Optimization | With Optimization |
| --------------- | -------------------- | ------------------- |
| First search | O(n) files | O(k) candidates |
| Repeated search | O(n) files | O(1) cache hit |
| Glob matching | O(n) regex tests | O(p) trie traversal |
| File list | O(n) filesystem | O(1) cache hit |
| File read | O(1) filesystem | O(1) memory |
Where n = total files, k = matching candidates, p = path depth.
## See Also
- [Tools Documentation](../tools/README.md) - Available MCP tools
- [Configuration Reference](../configuration.md) - All environment variables
