claude-self-reflect 2.6.0 → 2.7.1

package/.env.example

@@ -1,29 +1,30 @@
-# Claude Self-Reflect Environment Variables
-# Copy this file to .env and update with your values
-
-# Voyage AI API Key (optional - only needed for cloud mode)
-# Get your key at: https://www.voyageai.com/
-VOYAGE_KEY=your-voyage-api-key-here
-
 # Qdrant Configuration
 QDRANT_URL=http://localhost:6333
+QDRANT_PORT=6333
+QDRANT_MEMORY=4g
 
 # Embedding Configuration
-# Set to true to use local embeddings (default)
-# Set to false to use Voyage AI embeddings (requires VOYAGE_KEY)
+# Set to true to use local FastEmbed embeddings (recommended for privacy)
+# Set to false to use Voyage AI cloud embeddings
 PREFER_LOCAL_EMBEDDINGS=true
 
-# Memory Decay Configuration (optional)
+# Voyage AI Configuration (optional - only needed if PREFER_LOCAL_EMBEDDINGS=false)
+# Get your API key from https://www.voyageai.com/
+VOYAGE_KEY=
+
+# Memory Decay Configuration
 ENABLE_MEMORY_DECAY=true
 DECAY_WEIGHT=0.3
 DECAY_SCALE_DAYS=90
 
-# Import Configuration (optional)
-BATCH_SIZE=50
-CHUNK_SIZE=10
-WATCH_INTERVAL=60
+# Docker Configuration
+CLAUDE_LOGS_PATH=~/.claude/projects
+CONFIG_PATH=~/.claude-self-reflect/config
 
-# Docker Memory Limits
-# Increase if processing large conversation files
-WATCHER_MEM_LIMIT=2g
-QDRANT_MEM_LIMIT=1g
+# Performance Tuning
+WATCH_INTERVAL=5
+MAX_MEMORY_MB=600
+CHUNK_SIZE=50
+MAX_FILES_PER_CYCLE=10
+HOT_WINDOW_MINUTES=15
+MAX_COLD_FILES_PER_CYCLE=3

package/Dockerfile.importer

@@ -17,8 +17,12 @@ RUN pip install --no-cache-dir \
     voyageai==0.3.4 \
     tenacity==9.1.2
 
-# Note: The import script is mounted as a volume in docker-compose.yaml
-# This allows the container to work with both local development and global npm installs
+# Copy scripts into the image for npm package compatibility
+# This ensures scripts are available even without volume mounts
+COPY scripts/ /app/scripts/
+
+# Note: Volume mounts in docker-compose.yaml will override these files in local development
+# This provides compatibility for both local development and global npm installs
 
 # Default command (can be overridden by docker-compose)
 CMD ["python", "--version"]

package/Dockerfile.safe-watcher

@@ -0,0 +1,44 @@
+FROM python:3.12-slim
+
+# Install system dependencies
+RUN apt-get update && apt-get upgrade -y && apt-get install -y --no-install-recommends \
+    gcc \
+    g++ \
+    procps \
+    && rm -rf /var/lib/apt/lists/*
+
+# Upgrade pip and install setuptools
+RUN pip install --upgrade pip setuptools wheel
+
+# Install Python dependencies (matching importer)
+RUN pip install --no-cache-dir \
+    qdrant-client==1.15.0 \
+    fastembed==0.7.1 \
+    numpy==1.26.4 \
+    psutil==7.0.0 \
+    python-dotenv==1.0.0 \
+    voyageai==0.3.4 \
+    tenacity==9.1.2 \
+    humanize==4.12.3 \
+    tqdm==4.67.1
+
+# Pre-download and cache the FastEmbed model
+ENV FASTEMBED_CACHE_PATH=/root/.cache/fastembed
+RUN mkdir -p /root/.cache/fastembed && \
+    python -c "from fastembed import TextEmbedding; TextEmbedding(model_name='sentence-transformers/all-MiniLM-L6-v2')"
+
+# Set working directory
+WORKDIR /app
+
+# Copy scripts
+COPY scripts/ /scripts/
+
+# Create config directory
+RUN mkdir -p /config
+
+# Set environment variables for memory management
+ENV PYTHONUNBUFFERED=1
+ENV MALLOC_ARENA_MAX=2
+
+# Run the watcher loop
+CMD ["/scripts/watcher-loop.sh"]

package/README.md

@@ -1,5 +1,7 @@
 # Claude Self-Reflect
-
+<div align="center">
+![Alt](https://repobeats.axiom.co/api/embed/e45aa7276c6b2d1fbc46a9a3324e2231718787bb.svg "Repobeats analytics image")
+</div>
 <div align="center">
 
 [![npm version](https://badge.fury.io/js/claude-self-reflect.svg)](https://www.npmjs.com/package/claude-self-reflect)

package/docker-compose.yaml

@@ -8,7 +8,7 @@ services:
     command: chown -R 1000:1000 /config
     volumes:
       - ${CONFIG_PATH:-~/.claude-self-reflect/config}:/config
-    profiles: ["watch", "mcp", "import", "async"]
+    profiles: ["watch", "mcp", "import", "async", "safe-watch"]
 
   # Qdrant vector database - the heart of semantic search
   qdrant:
@@ -45,7 +45,7 @@ services:
       - OPENAI_API_KEY=${OPENAI_API_KEY:-}
       - VOYAGE_API_KEY=${VOYAGE_API_KEY:-}
       - VOYAGE_KEY=${VOYAGE_KEY:-}
-      - PREFER_LOCAL_EMBEDDINGS=${PREFER_LOCAL_EMBEDDINGS:-false}
+      - PREFER_LOCAL_EMBEDDINGS=${PREFER_LOCAL_EMBEDDINGS:-true}
       - EMBEDDING_MODEL=${EMBEDDING_MODEL:-voyage-3}
       - BATCH_SIZE=${BATCH_SIZE:-50}
       - CHUNK_SIZE=${CHUNK_SIZE:-10}
@@ -108,10 +108,10 @@ services:
       - MAX_CPU_PERCENT_PER_CORE=25  # 25% per core = 400% total on 16 cores
       - MAX_CONCURRENT_EMBEDDINGS=1  # Limit concurrent embeddings
       - MAX_CONCURRENT_QDRANT=2  # Limit concurrent Qdrant operations
-      - IMPORT_FREQUENCY=15  # Check every 15 seconds instead of 1
-      - BATCH_SIZE=3  # Process only 3 files at a time
-      - MEMORY_LIMIT_MB=600  # Memory limit (increased from 400MB for stability)
-      - MAX_QUEUE_SIZE=100  # Limit queue size
+      - IMPORT_FREQUENCY=5  # Check every 5 seconds for faster processing
+      - BATCH_SIZE=20  # Process 20 files at a time (increased from 3)
+      - MEMORY_LIMIT_MB=1500  # Memory limit (increased to 1.5GB)
+      - MAX_QUEUE_SIZE=500  # Increased queue size to handle backlog
       - MAX_BACKLOG_HOURS=24  # Alert if backlog > 24 hours
       - QDRANT_TIMEOUT=10  # 10 second timeout for Qdrant ops
       - MAX_RETRIES=3  # Retry failed operations
@@ -120,11 +120,11 @@ services:
       - LOGS_DIR=/logs
       - FASTEMBED_CACHE_PATH=/root/.cache/fastembed
       - MALLOC_ARENA_MAX=2  # MEMORY LEAK FIX: Limit glibc malloc arenas
-    restart: unless-stopped
-    profiles: ["watch"]
-    mem_limit: 500m
-    memswap_limit: 500m
-    cpus: 4.0  # Hard CPU limit: 4 cores max
+    restart: "no"  # DISABLED - causes system overload
+    profiles: ["watch-disabled"]  # Changed from "watch" to disable auto-start
+    mem_limit: 2g  # Increased from 500m to 2GB
+    memswap_limit: 2g
+    cpus: 8.0  # Increased CPU limit to 8 cores for faster processing
 
   # Async streaming importer - Ground-up async rewrite
   async-importer:
@@ -161,6 +161,38 @@ services:
     mem_limit: 4g
     memswap_limit: 4g
 
+  # Safe watcher with hot/warm/cold priority
+  safe-watcher:
+    build:
+      context: .
+      dockerfile: Dockerfile.safe-watcher
+    container_name: claude-reflection-safe-watcher
+    depends_on:
+      - init-permissions
+      - qdrant
+    volumes:
+      - ${CLAUDE_LOGS_PATH:-~/.claude/projects}:/logs:ro
+      - ${CONFIG_PATH:-~/.claude-self-reflect/config}:/config
+      - ./scripts:/scripts:ro
+    environment:
+      - QDRANT_URL=http://qdrant:6333
+      - STATE_FILE=/config/watcher-state.json
+      - VOYAGE_KEY=${VOYAGE_KEY:-}
+      - PREFER_LOCAL_EMBEDDINGS=${PREFER_LOCAL_EMBEDDINGS:-true}
+      - HOT_WINDOW_MINUTES=${HOT_WINDOW_MINUTES:-15}
+      - MAX_COLD_FILES_PER_CYCLE=${MAX_COLD_FILES_PER_CYCLE:-3}
+      - MAX_MEMORY_MB=${MAX_MEMORY_MB:-300}
+      - WATCH_INTERVAL_SECONDS=${WATCH_INTERVAL_SECONDS:-30}
+      - MAX_FILES_PER_CYCLE=${MAX_FILES_PER_CYCLE:-10}
+      - MAX_CHUNK_SIZE=${MAX_CHUNK_SIZE:-50}  # Messages per chunk for streaming
+      - PYTHONUNBUFFERED=1
+      - MALLOC_ARENA_MAX=2
+    restart: "no"  # Manual start only - prevent system overload
+    profiles: ["safe-watch"]  # Requires explicit profile to run
+    mem_limit: 600m  # Increased from 400m to handle large files safely
+    memswap_limit: 600m
+    cpus: 1.0  # Single CPU core limit
+
   # MCP server for Claude integration
   mcp-server:
     build:

package/mcp-server/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "claude-self-reflect-mcp"
-version = "2.5.19"
+version = "2.7.0"
 description = "MCP server for Claude self-reflection with memory decay"
 # readme = "README.md"
 requires-python = ">=3.10"