htm 0.0.31 → 0.0.32

data/docs/api/database.md

@@ -4,15 +4,13 @@ Database schema setup and configuration utilities for HTM.
 
 ## Overview
 
-`HTM::Database` provides class methods for setting up the HTM database schema, managing PostgreSQL connections, and configuring TimescaleDB hypertables.
+`HTM::Database` provides class methods for setting up the HTM database schema, managing PostgreSQL connections, and running migrations.
 
 **Key Features:**
 
 - Schema creation and migration
-- TimescaleDB hypertable setup
-- Extension verification (TimescaleDB, pgvector, pg_trgm)
+- Extension verification (pgvector, pg_trgm)
 - Connection configuration parsing
-- Automatic compression policies
 
 ## Class Definition
 
@@ -28,7 +26,7 @@ end
 
 ### `setup(db_url = nil)` {: #setup }
 
-Set up the HTM database schema and TimescaleDB hypertables.
+Set up the HTM database schema and run migrations.
 
 ```ruby
 HTM::Database.setup(db_url = nil)
@@ -52,10 +50,9 @@ HTM::Database.setup(db_url = nil)
 #### Side Effects
 
 - Connects to PostgreSQL database
-- Verifies required extensions (TimescaleDB, pgvector, pg_trgm)
+- Verifies required extensions (pgvector, pg_trgm)
 - Creates schema (tables, indexes, views)
-- Converts tables to hypertables
-- Sets up compression policies
+- Runs pending migrations
 - Prints status messages to stdout
 
 #### Examples
@@ -67,22 +64,18 @@ HTM::Database.setup
 # Use specific database URL
 HTM::Database.setup('postgresql://user:pass@host:5432/dbname')
 
-# Use TimescaleDB Cloud
-url = 'postgresql://tsdbadmin:pass@xxx.tsdb.cloud.timescale.com:37807/tsdb?sslmode=require'
+# Use a remote PostgreSQL database
+url = 'postgresql://user:pass@db.example.com:5432/htm_production?sslmode=require'
 HTM::Database.setup(url)
 ```
 
 #### Output
 
 ```
-✓ TimescaleDB version: 2.13.0
-✓ pgvector version: 0.5.1
+✓ pgvector version: 0.8.1+
 ✓ pg_trgm version: 1.6
 Creating HTM schema...
 ✓ Schema created
-✓ Created hypertable for operations_log
-✓ Created hypertable for nodes
-✓ Enabled compression for nodes older than 30 days
 ✓ HTM database schema created successfully
 ```
 
@@ -140,14 +133,14 @@ url = 'postgresql://user:pass@host:5432/db?sslmode=require'
 config = HTM::Database.parse_connection_url(url)
 # => { ..., sslmode: "require" }
 
-# TimescaleDB Cloud URL
-url = 'postgresql://tsdbadmin:secret@xxx.tsdb.cloud.timescale.com:37807/tsdb?sslmode=require'
+# Remote PostgreSQL URL
+url = 'postgresql://myuser:secret@db.example.com:5432/htm_production?sslmode=require'
 config = HTM::Database.parse_connection_url(url)
 # => {
-#   host: "xxx.tsdb.cloud.timescale.com",
-#   port: 37807,
-#   dbname: "tsdb",
-#   user: "tsdbadmin",
+#   host: "db.example.com",
+#   port: 5432,
+#   dbname: "htm_production",
+#   user: "myuser",
 #   password: "secret",
 #   sslmode: "require"
 # }
@@ -176,8 +169,8 @@ HTM::Database.parse_connection_params()
 
 | Variable | Description | Default |
 |----------|-------------|---------|
-| `HTM_DATABASE__HOST` | Database hostname | `'cw7rxj91bm.srbbwwxn56.tsdb.cloud.timescale.com'` |
-| `HTM_DATABASE__PORT` | Database port | `37807` |
+| `HTM_DATABASE__HOST` | Database hostname | `'localhost'` |
+| `HTM_DATABASE__PORT` | Database port | `5432` |
 | `HTM_DATABASE__NAME` | Database name | *required* |
 | `HTM_DATABASE__USER` | Database user | *required* |
 | `HTM_DATABASE__PASSWORD` | Database password | *required* |
@@ -186,18 +179,17 @@ HTM::Database.parse_connection_params()
 
 ```ruby
 # Set environment variables
-ENV['HTM_DATABASE__NAME'] = 'tsdb'
-ENV['HTM_DATABASE__USER'] = 'tsdbadmin'
+ENV['HTM_DATABASE__NAME'] = 'htm_development'
+ENV['HTM_DATABASE__USER'] = 'postgres'
 ENV['HTM_DATABASE__PASSWORD'] = 'secret'
 
 config = HTM::Database.parse_connection_params()
 # => {
-#   host: "cw7rxj91bm.srbbwwxn56.tsdb.cloud.timescale.com",
-#   port: 37807,
-#   dbname: "tsdb",
-#   user: "tsdbadmin",
-#   password: "secret",
-#   sslmode: "require"
+#   host: "localhost",
+#   port: 5432,
+#   dbname: "htm_development",
+#   user: "postgres",
+#   password: "secret"
 # }
 
 # Custom host and port
@@ -293,13 +285,9 @@ For detailed database schema documentation, see:
 ### 1. Verify Extensions
 
 ```ruby
-# Check TimescaleDB
-timescale = conn.exec("SELECT extversion FROM pg_extension WHERE extname='timescaledb'").first
-# => {"extversion"=>"2.13.0"}
-
 # Check pgvector
 pgvector = conn.exec("SELECT extversion FROM pg_extension WHERE extname='vector'").first
-# => {"extversion"=>"0.5.1"}
+# => {"extversion"=>"0.8.1"}
 
 # Check pg_trgm
 pg_trgm = conn.exec("SELECT extversion FROM pg_extension WHERE extname='pg_trgm'").first
@@ -335,33 +323,22 @@ conn.exec("SELECT add_compression_policy('nodes', INTERVAL '30 days', if_not_exi
 
 ## Environment Configuration
 
-### TimescaleDB Cloud
-
-Using URL (recommended):
+### Local PostgreSQL (macOS)
 
 ```bash
-# In ~/.bashrc__tiger
-export HTM_DATABASE__URL='postgresql://tsdbadmin:PASSWORD@SERVICE.tsdb.cloud.timescale.com:37807/tsdb?sslmode=require'
+export HTM_DATABASE__URL='postgresql://localhost/htm_development'
 ```
 
-Using individual variables:
+### Local PostgreSQL (with auth)
 
 ```bash
-# In ~/.bashrc__tiger
-export HTM_DATABASE__HOST='xxx.tsdb.cloud.timescale.com'
-export HTM_DATABASE__PORT=37807
-export HTM_DATABASE__NAME='tsdb'
-export HTM_DATABASE__USER='tsdbadmin'
-export HTM_DATABASE__PASSWORD='your_password'
+export HTM_DATABASE__URL='postgresql://user:pass@localhost:5432/htm_development'
 ```
 
-### Local PostgreSQL
+### Remote PostgreSQL (with SSL)
 
 ```bash
-export HTM_DATABASE__URL='postgresql://localhost/htm_dev'
-
-# Or with auth
-export HTM_DATABASE__URL='postgresql://user:pass@localhost:5432/htm_dev'
+export HTM_DATABASE__URL='postgresql://user:pass@db.example.com:5432/htm_production?sslmode=require'
 ```
 
 ### Docker PostgreSQL
@@ -461,7 +438,6 @@ htm = HTM.new(db_config: config)
 ### Extensions Not Available
 
 ```
-⚠ Warning: TimescaleDB extension not found
 ⚠ Warning: pgvector extension not found
 ```
 
@@ -469,12 +445,14 @@ htm = HTM.new(db_config: config)
 
 ```bash
 # Ubuntu/Debian
-sudo apt install postgresql-15-timescaledb postgresql-15-pgvector
+sudo apt install postgresql-17-pgvector
 
 # macOS with Homebrew
-brew install timescaledb pgvector
+brew install pgvector
 
-# Or use TimescaleDB Cloud (extensions pre-installed)
+# Then enable in your database:
+psql $HTM_DATABASE__URL -c "CREATE EXTENSION IF NOT EXISTS vector;"
+psql $HTM_DATABASE__URL -c "CREATE EXTENSION IF NOT EXISTS pg_trgm;"
 ```
 
 ### Connection Refused

data/docs/api/embedding-service.md

@@ -476,8 +476,8 @@ Search queries also generate embeddings:
 ```ruby
 # User searches for "database performance"
 results = htm.recall(
+  "database performance",
   timeframe: "last week",
-  topic: "database performance",
   strategy: :vector
 )
 

data/docs/api/index.md

@@ -33,11 +33,11 @@ HTM is a two-tier intelligent memory management system for LLM-based robots:
   <rect x="300" y="20" width="200" height="180" class="class-box" rx="5"/>
   <text x="400" y="45" text-anchor="middle" class="class-title">HTM</text>
   <line x1="310" y1="55" x2="490" y2="55" stroke="#4a9eff" stroke-width="1"/>
-  <text x="310" y="75" class="class-method">+ add_node()</text>
+  <text x="310" y="75" class="class-method">+ remember()</text>
   <text x="310" y="90" class="class-method">+ recall()</text>
-  <text x="310" y="105" class="class-method">+ retrieve()</text>
+  <text x="310" y="105" class="class-method">+ restore()</text>
   <text x="310" y="120" class="class-method">+ forget()</text>
-  <text x="310" y="135" class="class-method">+ create_context()</text>
+  <text x="310" y="135" class="class-method">+ working_memory()</text>
   <text x="310" y="150" class="class-method">+ memory_stats()</text>
   <text x="310" y="165" class="class-method">+ which_robot_said()</text>
   <text x="310" y="180" class="class-method">+ conversation_timeline()</text>
@@ -102,12 +102,23 @@ HTM is a two-tier intelligent memory management system for LLM-based robots:
 
 | Class | Purpose | Key Methods |
 |-------|---------|-------------|
-| [HTM](htm.md) | Main interface for memory management | `add_node`, `recall`, `retrieve`, `forget`, `create_context` |
+| [HTM](htm.md) | Main interface for memory management | `remember`, `recall`, `forget`, `restore`, `working_memory` |
 | [WorkingMemory](working-memory.md) | Token-limited active context | `add`, `evict_to_make_space`, `assemble_context` |
 | [LongTermMemory](long-term-memory.md) | Persistent PostgreSQL storage | `add`, `search`, `search_fulltext`, `search_hybrid` |
 | [EmbeddingService](embedding-service.md) | Vector embedding generation | `embed`, `count_tokens` |
 | [Database](database.md) | Schema setup and configuration | `setup`, `default_config` |
 
+### Sequel Models
+
+| Class | Table | Purpose |
+|-------|-------|---------|
+| `HTM::Models::Node` | nodes | Core memory storage |
+| `HTM::Models::Tag` | tags | Hierarchical tag names |
+| `HTM::Models::NodeTag` | node_tags | Node ↔ tag join |
+| `HTM::Models::NodeRelationship` | node_relationships | Weighted directed edges between nodes; Jaccard-similarity scores from tag co-occurrence |
+| `HTM::Models::Robot` | robots | Robot registry |
+| `HTM::Models::FileSource` | file_sources | Source file tracking |
+
 ### Common Usage Patterns
 
 #### Basic Memory Operations
@@ -117,14 +128,14 @@ HTM is a two-tier intelligent memory management system for LLM-based robots:
 htm = HTM.new(robot_name: "Assistant")
 
 # Add memories
-htm.add_node("fact_001", "PostgreSQL is our database",
-  type: :fact, importance: 7.0, tags: ["database"])
+htm.remember("PostgreSQL is our database",
+  type: :fact, tags: ["database"])
 
 # Recall memories
-memories = htm.recall(timeframe: "last week", topic: "PostgreSQL")
+memories = htm.recall("PostgreSQL", timeframe: "last week")
 
 # Create LLM context
-context = htm.create_context(strategy: :balanced)
+context = htm.working_memory.assemble_context(strategy: :balanced)
 ```
 
 #### Multi-Robot Collaboration
@@ -143,16 +154,16 @@ timeline = htm.conversation_timeline("deployment", limit: 20)
 ```ruby
 # Vector similarity search
 memories = htm.recall(
+  "API design decisions",
   timeframe: "last 30 days",
-  topic: "API design decisions",
   strategy: :vector,
   limit: 10
 )
 
 # Hybrid search (fulltext + vector)
 memories = htm.recall(
+  "security vulnerabilities",
   timeframe: "this month",
-  topic: "security vulnerabilities",
   strategy: :hybrid,
   limit: 20
 )
@@ -194,13 +205,13 @@ When adding nodes, you can specify a type:
 
 ## Context Assembly Strategies
 
-When creating context with `create_context`:
+When creating context with `working_memory.assemble_context`:
 
 | Strategy | Behavior |
 |----------|----------|
 | `:recent` | Most recently accessed first |
-| `:important` | Highest importance scores first |
-| `:balanced` | Weighted by importance and recency |
+| `:frequent` | Most frequently accessed first |
+| `:balanced` | Weighted by access frequency and recency |
 
 ## API Documentation
 
@@ -243,8 +254,8 @@ HTM is **not thread-safe** by default. Each instance maintains its own working m
 For large memory stores (>100K nodes):
 
 - Use hybrid search with appropriate `prefilter_limit`
-- Consider time-based partitioning (automatic with TimescaleDB)
-- Enable compression for old data (configured in schema)
+- Purge soft-deleted nodes periodically with `htm.purge_deleted`
+- Add partial indexes for frequently filtered columns
 
 ## Next Steps
 

data/docs/api/working-memory.md

@@ -297,7 +297,7 @@ assemble_context(strategy:, max_tokens: nil)
 
 | Parameter | Type | Default | Description |
 |-----------|------|---------|-------------|
-| `strategy` | Symbol | *required* | Assembly strategy (`:recent`, `:important`, `:balanced`) |
+| `strategy` | Symbol | *required* | Assembly strategy (`:recent`, `:frequent`, `:balanced`) |
 | `max_tokens` | Integer, nil | `@max_tokens` | Optional token limit |
 
 #### Assembly Strategies
@@ -309,10 +309,10 @@ assemble_context(strategy:, max_tokens: nil)
 # Good for: Conversational continuity
 ```
 
-**`:important`** - Highest importance scores first
+**`:frequent`** - Most frequently accessed first
 
 ```ruby
-# Ordered by importance score (highest first)
+# Ordered by access count (highest first)
 # Good for: Critical information prioritization
 ```
 
@@ -336,8 +336,8 @@ context = wm.assemble_context(strategy: :balanced)
 # Recent conversations only
 context = wm.assemble_context(strategy: :recent)
 
-# Most important information
-context = wm.assemble_context(strategy: :important)
+# Most frequently accessed information
+context = wm.assemble_context(strategy: :frequent)
 
 # Limited tokens
 context = wm.assemble_context(
@@ -524,8 +524,8 @@ class ContextManager
   end
 
   def for_analysis
-    # Important context for deep analysis
-    @wm.assemble_context(strategy: :important, max_tokens: 32000)
+    # Frequently accessed context for deep analysis
+    @wm.assemble_context(strategy: :frequent, max_tokens: 32000)
   end
 
   def for_general_task
@@ -652,7 +652,7 @@ end
 context = wm.assemble_context(strategy: :recent)
 
 # For analysis/reasoning
-context = wm.assemble_context(strategy: :important)
+context = wm.assemble_context(strategy: :frequent)
 
 # For general use
 context = wm.assemble_context(strategy: :balanced)

data/docs/architecture/index.md

@@ -37,7 +37,7 @@ HTM provides intelligent memory management through five core components that wor
   <rect x="300" y="380" width="200" height="120" fill="rgba(244, 67, 54, 0.2)" stroke="#F44336" stroke-width="2" rx="5"/>
   <text x="400" y="415" text-anchor="middle" fill="#E0E0E0" font-size="14" font-weight="bold">Database</text>
   <text x="400" y="435" text-anchor="middle" fill="#B0B0B0" font-size="11">PostgreSQL 16+</text>
-  <text x="400" y="455" text-anchor="middle" fill="#B0B0B0" font-size="11">TimescaleDB</text>
+  <text x="400" y="455" text-anchor="middle" fill="#B0B0B0" font-size="11">Sequel ORM</text>
   <text x="400" y="475" text-anchor="middle" fill="#B0B0B0" font-size="11">pgvector + pg_trgm</text>
 
   <!-- Connections -->
@@ -102,7 +102,7 @@ Durable PostgreSQL storage for permanent knowledge retention. All memories are s
 **Characteristics:**
 
 - **Capacity:** Effectively unlimited
-- **Storage:** PostgreSQL with TimescaleDB extension
+- **Storage:** PostgreSQL with pgvector and pg_trgm extensions
 - **Retention:** Permanent (explicit deletion only)
 - **Access Pattern:** RAG-based retrieval (semantic + temporal)
 - **Lifetime:** Forever
@@ -126,11 +126,10 @@ Generates vector embeddings for semantic search and manages token counting for m
 
 ### Database
 
-PostgreSQL 16+ with extensions for time-series optimization, vector similarity search, and full-text search.
+PostgreSQL 16+ with extensions for vector similarity search, full-text search, and fuzzy matching.
 
 **Key Extensions:**
 
-- **TimescaleDB**: Hypertable partitioning, compression policies, time-range optimization
 - **pgvector**: Vector similarity search with HNSW indexing
 - **pg_trgm**: Trigram-based fuzzy text matching
 
@@ -282,8 +281,7 @@ Explore detailed architecture documentation:
 | Layer | Technology | Purpose |
 |-------|-----------|---------|
 | **Language** | Ruby 3.2+ | Core implementation |
-| **Database** | PostgreSQL 16+ | Relational storage |
-| **Time-Series** | TimescaleDB | Hypertable partitioning, compression |
+| **Database** | PostgreSQL 16+ | Relational storage with soft delete |
 | **Vector Search** | pgvector | Semantic similarity (HNSW) |
 | **Full-Text** | pg_trgm | Fuzzy text matching |
 | **Embeddings** | RubyLLM (multi-provider) | Vector generation |
@@ -334,7 +332,7 @@ Explore detailed architecture documentation:
 
 ## Related Documentation
 
-- [Installation Guide](../getting-started/installation.md) - Setup PostgreSQL, TimescaleDB, and dependencies
+- [Installation Guide](../getting-started/installation.md) - Setup PostgreSQL and dependencies
 - [Quick Start](../getting-started/quick-start.md) - Get started with HTM in 5 minutes
 - [API Reference](../api/htm.md) - Complete API documentation
 - [Architecture Decision Records](adrs/index.md) - Detailed decision history

data/docs/architecture/overview.md

@@ -79,9 +79,9 @@ HTM implements a layered architecture with clear separation of concerns between
 #### Data Layer
 
 - **PostgreSQL**: Relational storage with ACID guarantees
-- **TimescaleDB**: Time-series optimization and compression
 - **pgvector**: Vector similarity search with HNSW
 - **pg_trgm**: Fuzzy text matching for search
+- **Sequel**: Ruby ORM for database access
 
 ## Component Diagrams
 
@@ -238,7 +238,6 @@ Audit trail of all memory operations for debugging and replay.
 |-----------|---------|---------|------------|
 | **Ruby** | 3.2+ | Implementation language | Readable, expressive, mature ecosystem |
 | **PostgreSQL** | 16+ | Relational database | ACID guarantees, rich extensions, production-proven |
-| **TimescaleDB** | 2.13+ | Time-series extension | Hypertable partitioning, automatic compression |
 | **pgvector** | 0.5+ | Vector similarity | HNSW indexing, PostgreSQL-native, fast approximate search |
 | **pg_trgm** | - | Fuzzy text search | Built-in PostgreSQL extension for trigram matching |
 
@@ -284,8 +283,8 @@ Based on typical production workloads with 10,000 nodes in long-term memory (cli
 | `recall()` (vector) | 80ms | 140ms | 230ms | Client-side query embedding + vector search |
 | `recall()` (fulltext) | 30ms | 60ms | 100ms | GIN index search (no embedding needed) |
 | `recall()` (hybrid) | 110ms | 190ms | 330ms | Client-side embedding + hybrid search |
-| `retrieve()` | 5ms | 10ms | 20ms | Simple primary key lookup |
-| `create_context()` | 8ms | 15ms | 25ms | In-memory sort + join |
+| `HTM::Models::Node.first()` | 5ms | 10ms | 20ms | Simple primary key lookup |
+| `working_memory.assemble_context()` | 8ms | 15ms | 25ms | In-memory sort + join |
 | `forget()` | 10ms | 20ms | 40ms | DELETE with cascades |
 
 !!! tip "Performance Optimization"
@@ -308,15 +307,13 @@ Based on typical production workloads with 10,000 nodes in long-term memory (cli
 | Component | Size Estimate | Compression |
 |-----------|--------------|-------------|
 | Node (text only) | ~1KB average | None |
-| Node (with embedding) | ~7KB (1536 dims × 4 bytes) | TimescaleDB compression (70-90%) |
+| Node (with embedding) | ~7KB (1536 dims × 4 bytes) | TOAST (auto) |
 | Indexes | ~2x data size | Minimal |
-| Operations log | ~200 bytes/op | TimescaleDB compression |
 
 **Example:** 100,000 nodes with embeddings:
 
 - Raw data: ~700 MB
 - With indexes: ~2.1 GB
-- With compression (after 30 days): ~300 MB
 
 ## Scalability Considerations
 
@@ -354,7 +351,7 @@ Based on typical production workloads with 10,000 nodes in long-term memory (cli
 - Add PostgreSQL read replicas
 - Route `recall()` and `retrieve()` to replicas
 - Primary handles writes only
-- TimescaleDB native replication support
+- PostgreSQL native replication support
 
 !!! warning "Consistency Considerations"
     Read replicas may lag primary by seconds. For strong consistency requirements, query primary database.

data/docs/assets/images/htm-architecture-overview.svg

@@ -13,7 +13,7 @@
   <!-- HTM API -->
   <rect x="200" y="190" width="500" height="80" fill="rgba(33, 150, 243, 0.3)" stroke="#2196F3" stroke-width="3" rx="8"/>
   <text x="450" y="225" text-anchor="middle" fill="#FFFFFF" font-size="16" font-weight="bold">HTM API</text>
-  <text x="450" y="250" text-anchor="middle" fill="#B0B0B0" font-size="13">add_node(), recall(), create_context(), etc.</text>
+  <text x="450" y="250" text-anchor="middle" fill="#B0B0B0" font-size="13">remember(), recall(), forget(), restore(), etc.</text>
 
   <!-- Arrows down to both memories -->
   <path d="M 320 270 L 320 330" stroke="#2196F3" stroke-width="3" fill="none" marker-end="url(#arrow1)"/>

data/docs/assets/images/htm-context-assembly-flow.svg

@@ -4,7 +4,7 @@
 
   <!-- User -->
   <rect x="50" y="80" width="200" height="60" fill="rgba(76, 175, 80, 0.3)" stroke="#4CAF50" stroke-width="3" rx="8"/>
-  <text x="150" y="110" text-anchor="middle" fill="#FFFFFF" font-size="13" font-weight="bold">User: create_context</text>
+  <text x="150" y="110" text-anchor="middle" fill="#FFFFFF" font-size="13" font-weight="bold">User: assemble_context</text>
   <text x="150" y="128" text-anchor="middle" fill="#B0B0B0" font-size="11">(with strategy)</text>
 
   <!-- HTM -->
@@ -81,7 +81,7 @@
   <text x="695" y="165" text-anchor="middle" fill="#673AB7" font-size="10">:recent</text>
 
   <path d="M 670 210 L 720 210 L 720 250" stroke="#F44336" stroke-width="2" fill="none" marker-end="url(#arrowR)"/>
-  <text x="695" y="215" text-anchor="middle" fill="#F44336" font-size="10">:important</text>
+  <text x="695" y="215" text-anchor="middle" fill="#F44336" font-size="10">:frequent</text>
 
   <path d="M 670 245 L 720 245 L 720 330" stroke="#4CAF50" stroke-width="2" fill="none" marker-end="url(#arrowG)"/>
   <text x="695" y="260" text-anchor="middle" fill="#4CAF50" font-size="10">:balanced</text>

data/docs/assets/images/htm-layered-architecture.svg

@@ -5,11 +5,11 @@
   <!-- Layer 1: API Layer -->
   <rect x="100" y="60" width="600" height="100" fill="rgba(76, 175, 80, 0.2)" stroke="#4CAF50" stroke-width="2" rx="5"/>
   <text x="400" y="85" text-anchor="middle" fill="#E0E0E0" font-size="16" font-weight="bold">API Layer</text>
-  <text x="200" y="110" text-anchor="middle" fill="#B0B0B0" font-size="12">add_node()</text>
+  <text x="200" y="110" text-anchor="middle" fill="#B0B0B0" font-size="12">remember()</text>
   <text x="320" y="110" text-anchor="middle" fill="#B0B0B0" font-size="12">recall()</text>
-  <text x="440" y="110" text-anchor="middle" fill="#B0B0B0" font-size="12">retrieve()</text>
+  <text x="440" y="110" text-anchor="middle" fill="#B0B0B0" font-size="12">restore()</text>
   <text x="560" y="110" text-anchor="middle" fill="#B0B0B0" font-size="12">forget()</text>
-  <text x="400" y="135" text-anchor="middle" fill="#B0B0B0" font-size="12">create_context() • memory_stats() • which_robot_said()</text>
+  <text x="400" y="135" text-anchor="middle" fill="#B0B0B0" font-size="12">assemble_context() • memory_stats() • which_robot_said()</text>
 
   <!-- Layer 2: Coordination Layer -->
   <rect x="100" y="180" width="600" height="100" fill="rgba(33, 150, 243, 0.2)" stroke="#2196F3" stroke-width="2" rx="5"/>

data/docs/assets/images/two-tier-memory-architecture.svg

@@ -16,7 +16,7 @@
   <rect x="450" y="80" width="300" height="180" fill="rgba(156, 39, 176, 0.2)" stroke="#9C27B0" stroke-width="3" rx="5"/>
   <text x="600" y="110" text-anchor="middle" fill="#E0E0E0" font-size="16" font-weight="bold">Long-Term Memory (Cold)</text>
   <text x="480" y="140" fill="#B0B0B0" font-size="12">Capacity: Unlimited</text>
-  <text x="480" y="160" fill="#B0B0B0" font-size="12">Storage: PostgreSQL + TimescaleDB</text>
+  <text x="480" y="160" fill="#B0B0B0" font-size="12">Storage: PostgreSQL + pgvector</text>
   <text x="480" y="180" fill="#B0B0B0" font-size="12">Speed: O(log n) with indexes</text>
   <text x="480" y="200" fill="#B0B0B0" font-size="12">Lifetime: Permanent</text>
   <text x="480" y="220" fill="#B0B0B0" font-size="12">Retrieval: RAG (semantic + temporal)</text>

data/docs/database/README.md

@@ -5,6 +5,7 @@
 | Name | Columns | Comment | Type |
 | ---- | ------- | ------- | ---- |
 | [public.file_sources](public.file_sources.md) | 9 | Source file metadata for loaded documents | BASE TABLE |
+| [public.node_relationships](public.node_relationships.md) | 8 | Weighted directed edges between related nodes (Jaccard similarity) | BASE TABLE |
 | [public.node_tags](public.node_tags.md) | 5 | Join table connecting nodes to tags (many-to-many) | BASE TABLE |
 | [public.nodes](public.nodes.md) | 14 | Core memory storage for conversation messages and context | BASE TABLE |
 | [public.robot_nodes](public.robot_nodes.md) | 10 | Join table connecting robots to nodes (many-to-many) | BASE TABLE |