agent-cli 0.72.3__tar.gz → 0.72.5__tar.gz

{agent_cli-0.72.3 → agent_cli-0.72.5}/.claude-plugin/README.md

@@ -30,7 +30,7 @@ uvx --python 3.13 agent-cli dev new my-feature --agent --prompt "..."
 claude plugin marketplace add basnijholt/agent-cli
 
 # Then install
-claude plugin install agent-cli@agent-cli-dev
+claude plugin install agent-cli-dev@agent-cli
 ```
 
 ## Usage

{agent_cli-0.72.3 → agent_cli-0.72.5}/.github/workflows/docker.yml

@@ -104,6 +104,10 @@ jobs:
         include:
           - image: transcribe-proxy
             dockerfile: docker/transcribe-proxy.Dockerfile
+          - image: rag-proxy
+            dockerfile: docker/rag-proxy.Dockerfile
+          - image: memory-proxy
+            dockerfile: docker/memory-proxy.Dockerfile
 
     steps:
       - name: Checkout repository

{agent_cli-0.72.3 → agent_cli-0.72.5}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: agent-cli
-Version: 0.72.3
+Version: 0.72.5
 Summary: A suite of AI-powered command-line tools for text correction, audio transcription, and voice assistance.
 Project-URL: Homepage, https://github.com/basnijholt/agent-cli
 Author-email: Bas Nijholt <bas@nijho.lt>
@@ -88,7 +88,7 @@ Description-Content-Type: text/markdown
 
 # Agent CLI
 
-<img src="https://raw.githubusercontent.com/basnijholt/agent-cli/refs/heads/main/.github/logo.svg" alt="agent-cli logo" align="right" style="width: 250px;" />
+<img src="https://raw.githubusercontent.com/basnijholt/agent-cli/refs/heads/main/.github/logo.svg" alt="agent-cli logo" align="right" style="width: 200px;" />
 
 `agent-cli` is a collection of **_local-first_**, AI-powered command-line agents that run entirely on your machine.
 It provides a suite of powerful tools for voice and text interaction, designed for privacy, offline capability, and seamless integration with system-wide hotkeys and workflows.
@@ -391,7 +391,7 @@ agent-cli dev install-skill
 
 # Option 2: Install via Claude Code plugin marketplace
 claude plugin marketplace add basnijholt/agent-cli
-claude plugin install agent-cli@agent-cli-dev
+claude plugin install agent-cli-dev@agent-cli
 ```
 
 Once installed, Claude Code can automatically use this skill when you ask to:

{agent_cli-0.72.3 → agent_cli-0.72.5}/README.md

@@ -1,6 +1,6 @@
 # Agent CLI
 
-<img src="https://raw.githubusercontent.com/basnijholt/agent-cli/refs/heads/main/.github/logo.svg" alt="agent-cli logo" align="right" style="width: 250px;" />
+<img src="https://raw.githubusercontent.com/basnijholt/agent-cli/refs/heads/main/.github/logo.svg" alt="agent-cli logo" align="right" style="width: 200px;" />
 
 `agent-cli` is a collection of **_local-first_**, AI-powered command-line agents that run entirely on your machine.
 It provides a suite of powerful tools for voice and text interaction, designed for privacy, offline capability, and seamless integration with system-wide hotkeys and workflows.
@@ -303,7 +303,7 @@ agent-cli dev install-skill
 
 # Option 2: Install via Claude Code plugin marketplace
 claude plugin marketplace add basnijholt/agent-cli
-claude plugin install agent-cli@agent-cli-dev
+claude plugin install agent-cli-dev@agent-cli
 ```
 
 Once installed, Claude Code can automatically use this skill when you ask to:

{agent_cli-0.72.3 → agent_cli-0.72.5}/agent_cli/dev/cli.py

@@ -176,6 +176,8 @@ def _info(msg: str) -> None:
     # Style commands (messages starting with "Running: ")
     if msg.startswith("Running: "):
         cmd = msg[9:]  # Remove "Running: " prefix
+        # Escape brackets to prevent Rich from interpreting them as markup
+        cmd = cmd.replace("[", r"\[")
         console.print(f"[dim]→[/dim] Running: [bold cyan]{cmd}[/bold cyan]")
     else:
         console.print(f"[dim]→[/dim] {msg}")
@@ -1777,7 +1779,9 @@ def install_skill(
     - Works when you ask to "work on multiple features" or "parallelize tasks"
 
     **Alternative:** Install globally via Claude Code plugin marketplace:
-    `claude plugin marketplace add basnijholt/agent-cli`
+
+    1. `claude plugin marketplace add basnijholt/agent-cli`
+    2. `claude plugin install agent-cli-dev@agent-cli`
     """
     # Use current repo root (works in worktrees too)
     repo_root = _get_current_repo_root()

{agent_cli-0.72.3 → agent_cli-0.72.5}/agent_cli/install/extras.py

@@ -68,7 +68,9 @@ def _install_via_uv_tool(extras: list[str], *, quiet: bool = False) -> bool:
     if quiet:
         cmd.append("-q")
     # Use stderr for status messages so they don't pollute stdout (e.g., for hotkey notifications)
-    err_console.print(f"Running: [cyan]{' '.join(cmd)}[/]")
+    # Escape brackets to prevent Rich from interpreting them as markup
+    cmd_str = " ".join(cmd).replace("[", r"\[")
+    err_console.print(f"Running: [cyan]{cmd_str}[/]")
     result = subprocess.run(cmd, check=False)
     return result.returncode == 0
 

{agent_cli-0.72.3 → agent_cli-0.72.5}/docker/docker-compose.yml

@@ -41,6 +41,17 @@
 #   LLM_OPENAI_MODEL   - Model name for OpenAI LLM provider
 #   OPENAI_BASE_URL    - OpenAI-compatible LLM endpoint
 #   OPENAI_API_KEY     - API key for OpenAI LLM
+#
+#   RAG_PORT           - RAG proxy port (default: 8000)
+#   RAG_LIMIT          - Number of document chunks per query (default: 3)
+#   RAG_ENABLE_TOOLS   - Enable read_full_document tool (default: true)
+#   EMBEDDING_MODEL    - Embedding model for RAG/memory (default: text-embedding-3-small)
+#
+#   MEMORY_PORT        - Memory proxy port (default: 8100)
+#   MEMORY_TOP_K       - Number of memories per query (default: 5)
+#   MEMORY_MAX_ENTRIES - Max entries per conversation (default: 500)
+#   MEMORY_SUMMARIZATION - Enable fact extraction (default: true)
+#   MEMORY_GIT_VERSIONING - Enable git versioning (default: true)
 
 services:
   # ===========================================================================
@@ -227,6 +238,69 @@ services:
       # - OPENAI_API_KEY=your-api-key
     restart: unless-stopped
 
+  # ===========================================================================
+  # RAG Proxy
+  # ===========================================================================
+
+  rag-proxy:
+    image: ghcr.io/basnijholt/agent-cli-rag-proxy:latest
+    build:
+      context: ..
+      dockerfile: docker/rag-proxy.Dockerfile
+    profiles: [cuda, cpu]
+    ports:
+      - "8000:8000"
+    volumes:
+      - rag-docs:/data/docs
+      - rag-db:/data/db
+      - rag-cache:/home/rag/.cache
+    environment:
+      - RAG_HOST=0.0.0.0
+      - RAG_PORT=${RAG_PORT:-8000}
+      - RAG_DOCS_FOLDER=/data/docs
+      - RAG_CHROMA_PATH=/data/db
+      - RAG_LIMIT=${RAG_LIMIT:-3}
+      - RAG_ENABLE_TOOLS=${RAG_ENABLE_TOOLS:-true}
+      - EMBEDDING_MODEL=${EMBEDDING_MODEL:-text-embedding-3-small}
+      - LOG_LEVEL=${LOG_LEVEL:-info}
+      # OpenAI-compatible backend configuration
+      - OPENAI_BASE_URL=${OPENAI_BASE_URL:-http://ollama:11434/v1}
+      - OPENAI_API_KEY=${OPENAI_API_KEY:-}
+    restart: unless-stopped
+
+  # ===========================================================================
+  # Memory Proxy
+  # ===========================================================================
+
+  memory-proxy:
+    image: ghcr.io/basnijholt/agent-cli-memory-proxy:latest
+    build:
+      context: ..
+      dockerfile: docker/memory-proxy.Dockerfile
+    profiles: [cuda, cpu]
+    ports:
+      - "8100:8100"
+    volumes:
+      - memory-data:/data/memory
+      - memory-cache:/home/memory/.cache
+    environment:
+      - MEMORY_HOST=0.0.0.0
+      - MEMORY_PORT=${MEMORY_PORT:-8100}
+      - MEMORY_PATH=/data/memory
+      - MEMORY_TOP_K=${MEMORY_TOP_K:-5}
+      - MEMORY_MAX_ENTRIES=${MEMORY_MAX_ENTRIES:-500}
+      - MEMORY_MMR_LAMBDA=${MEMORY_MMR_LAMBDA:-0.7}
+      - MEMORY_RECENCY_WEIGHT=${MEMORY_RECENCY_WEIGHT:-0.2}
+      - MEMORY_SCORE_THRESHOLD=${MEMORY_SCORE_THRESHOLD:-0.35}
+      - MEMORY_SUMMARIZATION=${MEMORY_SUMMARIZATION:-true}
+      - MEMORY_GIT_VERSIONING=${MEMORY_GIT_VERSIONING:-true}
+      - EMBEDDING_MODEL=${EMBEDDING_MODEL:-text-embedding-3-small}
+      - LOG_LEVEL=${LOG_LEVEL:-info}
+      # OpenAI-compatible backend configuration
+      - OPENAI_BASE_URL=${OPENAI_BASE_URL:-http://ollama:11434/v1}
+      - OPENAI_API_KEY=${OPENAI_API_KEY:-}
+    restart: unless-stopped
+
 volumes:
   ollama-data:
     name: agent-cli-ollama-data
@@ -236,3 +310,13 @@ volumes:
     name: agent-cli-whisper-cache
   tts-cache:
     name: agent-cli-tts-cache
+  rag-docs:
+    name: agent-cli-rag-docs
+  rag-db:
+    name: agent-cli-rag-db
+  rag-cache:
+    name: agent-cli-rag-cache
+  memory-data:
+    name: agent-cli-memory-data
+  memory-cache:
+    name: agent-cli-memory-cache

agent_cli-0.72.5/docker/memory-proxy.Dockerfile

@@ -0,0 +1,121 @@
+# Dockerfile for agent-cli Memory Proxy server
+# A long-term memory proxy that stores facts and injects relevant context into LLM requests
+#
+# Build example:
+#   docker build -f docker/memory-proxy.Dockerfile -t agent-cli-memory-proxy .
+#
+# Run examples:
+#   docker run -p 8100:8100 -v ./memory:/data/memory agent-cli-memory-proxy
+#
+#   # With custom OpenAI-compatible backend:
+#   docker run -p 8100:8100 \
+#     -v ./memory:/data/memory \
+#     -e OPENAI_BASE_URL=http://ollama:11434/v1 \
+#     agent-cli-memory-proxy
+#
+# Environment variables (priority: env var > config file > default):
+#   MEMORY_PATH          - Directory for memory storage (default: /data/memory)
+#   MEMORY_TOP_K         - Number of memories to retrieve per query (default: 5)
+#   MEMORY_MAX_ENTRIES   - Max entries per conversation before eviction (default: 500)
+#   MEMORY_MMR_LAMBDA    - MMR lambda 0-1, higher=relevance (default: 0.7)
+#   MEMORY_RECENCY_WEIGHT - Weight for recency vs semantic (default: 0.2)
+#   MEMORY_SCORE_THRESHOLD - Min relevance threshold (default: 0.35)
+#   MEMORY_SUMMARIZATION - Enable fact extraction: true/false (default: true)
+#   MEMORY_GIT_VERSIONING - Enable git versioning: true/false (default: true)
+#   EMBEDDING_MODEL      - Embedding model name (default: text-embedding-3-small)
+#   OPENAI_BASE_URL      - OpenAI-compatible API base URL
+#   OPENAI_API_KEY       - API key for embeddings and chat
+#   MEMORY_HOST          - Server bind address (default: 0.0.0.0)
+#   MEMORY_PORT          - Server port (default: 8100)
+#   LOG_LEVEL            - Logging level: debug, info, warning, error (default: info)
+
+# =============================================================================
+# Builder stage - install dependencies and project
+# =============================================================================
+FROM python:3.13-slim AS builder
+
+RUN apt-get update && \
+    apt-get install -y --no-install-recommends git build-essential && \
+    apt-get clean && rm -rf /var/lib/apt/lists/*
+
+COPY --from=ghcr.io/astral-sh/uv:latest /uv /uvx /bin/
+
+WORKDIR /app
+
+COPY pyproject.toml uv.lock README.md ./
+COPY .git ./.git
+COPY agent_cli ./agent_cli
+COPY scripts ./scripts
+RUN uv sync --frozen --no-dev --no-editable --extra memory
+
+# =============================================================================
+# Runtime stage - minimal image
+# =============================================================================
+FROM debian:bookworm-slim
+
+COPY --from=ghcr.io/astral-sh/uv:latest /uv /uvx /bin/
+
+# Install runtime dependencies:
+# - libgomp1: Required by onnxruntime for parallel processing
+# - git: Required for memory git versioning feature
+RUN apt-get update && \
+    apt-get install -y --no-install-recommends libgomp1 git && \
+    apt-get clean && rm -rf /var/lib/apt/lists/*
+
+ENV UV_PYTHON_INSTALL_DIR=/opt/python
+RUN uv python install 3.13
+
+RUN groupadd -g 1000 memory && useradd -m -u 1000 -g memory memory
+
+WORKDIR /app
+
+COPY --from=builder /app/.venv /app/.venv
+
+RUN ln -sf $(uv python find 3.13) /app/.venv/bin/python && \
+    ln -s /app/.venv/bin/agent-cli /usr/local/bin/agent-cli
+
+# Create data directory
+RUN mkdir -p /data/memory && chown -R memory:memory /data
+
+USER memory
+
+# Configure git for memory versioning (required for commits)
+RUN git config --global user.email "memory-proxy@agent-cli.local" && \
+    git config --global user.name "Memory Proxy"
+
+# Cache directory for models (embeddings)
+ENV HF_HOME=/home/memory/.cache/huggingface
+RUN mkdir -p /home/memory/.cache/huggingface
+
+EXPOSE 8100
+
+ENV MEMORY_HOST=0.0.0.0 \
+    MEMORY_PORT=8100 \
+    MEMORY_PATH=/data/memory \
+    MEMORY_TOP_K=5 \
+    MEMORY_MAX_ENTRIES=500 \
+    MEMORY_MMR_LAMBDA=0.7 \
+    MEMORY_RECENCY_WEIGHT=0.2 \
+    MEMORY_SCORE_THRESHOLD=0.35 \
+    MEMORY_SUMMARIZATION=true \
+    MEMORY_GIT_VERSIONING=true \
+    EMBEDDING_MODEL=text-embedding-3-small \
+    LOG_LEVEL=info
+
+HEALTHCHECK --interval=30s --timeout=10s --start-period=30s --retries=3 \
+    CMD /app/.venv/bin/python -c "import urllib.request; urllib.request.urlopen('http://localhost:${MEMORY_PORT}/health')" || exit 1
+
+ENTRYPOINT ["sh", "-c", "agent-cli memory proxy \
+    --host ${MEMORY_HOST} \
+    --port ${MEMORY_PORT} \
+    --memory-path ${MEMORY_PATH} \
+    --default-top-k ${MEMORY_TOP_K} \
+    --max-entries ${MEMORY_MAX_ENTRIES} \
+    --mmr-lambda ${MEMORY_MMR_LAMBDA} \
+    --recency-weight ${MEMORY_RECENCY_WEIGHT} \
+    --score-threshold ${MEMORY_SCORE_THRESHOLD} \
+    --embedding-model ${EMBEDDING_MODEL} \
+    --log-level ${LOG_LEVEL} \
+    $([ \"${MEMORY_SUMMARIZATION}\" = \"false\" ] && echo '--no-summarization' || echo '--summarization') \
+    $([ \"${MEMORY_GIT_VERSIONING}\" = \"false\" ] && echo '--no-git-versioning' || echo '--git-versioning') \
+    ${MEMORY_EXTRA_ARGS:-}"]

agent_cli-0.72.5/docker/rag-proxy.Dockerfile

@@ -0,0 +1,106 @@
+# Dockerfile for agent-cli RAG Proxy server
+# A document-aware proxy that indexes files and injects relevant context into LLM requests
+#
+# Build example:
+#   docker build -f docker/rag-proxy.Dockerfile -t agent-cli-rag-proxy .
+#
+# Run examples:
+#   docker run -p 8000:8000 -v ./docs:/data/docs agent-cli-rag-proxy
+#
+#   # With custom OpenAI-compatible backend:
+#   docker run -p 8000:8000 \
+#     -v ./docs:/data/docs \
+#     -v ./rag_db:/data/db \
+#     -e OPENAI_BASE_URL=http://ollama:11434/v1 \
+#     agent-cli-rag-proxy
+#
+# Environment variables (priority: env var > config file > default):
+#   RAG_DOCS_FOLDER      - Folder to watch for documents (default: /data/docs)
+#   RAG_CHROMA_PATH      - ChromaDB storage directory (default: /data/db)
+#   RAG_LIMIT            - Number of chunks to retrieve per query (default: 3)
+#   RAG_ENABLE_TOOLS     - Enable read_full_document tool: true/false (default: true)
+#   EMBEDDING_MODEL      - Embedding model name (default: text-embedding-3-small)
+#   OPENAI_BASE_URL      - OpenAI-compatible API base URL
+#   OPENAI_API_KEY       - API key for embeddings and chat
+#   RAG_HOST             - Server bind address (default: 0.0.0.0)
+#   RAG_PORT             - Server port (default: 8000)
+#   LOG_LEVEL            - Logging level: debug, info, warning, error (default: info)
+
+# =============================================================================
+# Builder stage - install dependencies and project
+# =============================================================================
+FROM python:3.13-slim AS builder
+
+RUN apt-get update && \
+    apt-get install -y --no-install-recommends git build-essential && \
+    apt-get clean && rm -rf /var/lib/apt/lists/*
+
+COPY --from=ghcr.io/astral-sh/uv:latest /uv /uvx /bin/
+
+WORKDIR /app
+
+COPY pyproject.toml uv.lock README.md ./
+COPY .git ./.git
+COPY agent_cli ./agent_cli
+COPY scripts ./scripts
+RUN uv sync --frozen --no-dev --no-editable --extra rag
+
+# =============================================================================
+# Runtime stage - minimal image
+# =============================================================================
+FROM debian:bookworm-slim
+
+COPY --from=ghcr.io/astral-sh/uv:latest /uv /uvx /bin/
+
+# Install runtime dependencies:
+# - libgomp1: Required by onnxruntime for parallel processing
+# - git: Required for .git detection in some document types
+RUN apt-get update && \
+    apt-get install -y --no-install-recommends libgomp1 git && \
+    apt-get clean && rm -rf /var/lib/apt/lists/*
+
+ENV UV_PYTHON_INSTALL_DIR=/opt/python
+RUN uv python install 3.13
+
+RUN groupadd -g 1000 rag && useradd -m -u 1000 -g rag rag
+
+WORKDIR /app
+
+COPY --from=builder /app/.venv /app/.venv
+
+RUN ln -sf $(uv python find 3.13) /app/.venv/bin/python && \
+    ln -s /app/.venv/bin/agent-cli /usr/local/bin/agent-cli
+
+# Create data directories
+RUN mkdir -p /data/docs /data/db && chown -R rag:rag /data
+
+USER rag
+
+# Cache directory for models (embeddings, reranker)
+ENV HF_HOME=/home/rag/.cache/huggingface
+RUN mkdir -p /home/rag/.cache/huggingface
+
+EXPOSE 8000
+
+ENV RAG_HOST=0.0.0.0 \
+    RAG_PORT=8000 \
+    RAG_DOCS_FOLDER=/data/docs \
+    RAG_CHROMA_PATH=/data/db \
+    RAG_LIMIT=3 \
+    RAG_ENABLE_TOOLS=true \
+    EMBEDDING_MODEL=text-embedding-3-small \
+    LOG_LEVEL=info
+
+HEALTHCHECK --interval=30s --timeout=10s --start-period=30s --retries=3 \
+    CMD /app/.venv/bin/python -c "import urllib.request; urllib.request.urlopen('http://localhost:${RAG_PORT}/health')" || exit 1
+
+ENTRYPOINT ["sh", "-c", "agent-cli rag-proxy \
+    --host ${RAG_HOST} \
+    --port ${RAG_PORT} \
+    --docs-folder ${RAG_DOCS_FOLDER} \
+    --chroma-path ${RAG_CHROMA_PATH} \
+    --limit ${RAG_LIMIT} \
+    --embedding-model ${EMBEDDING_MODEL} \
+    --log-level ${LOG_LEVEL} \
+    $([ \"${RAG_ENABLE_TOOLS}\" = \"false\" ] && echo '--no-rag-tools' || echo '--rag-tools') \
+    ${RAG_EXTRA_ARGS:-}"]

{agent_cli-0.72.3 → agent_cli-0.72.5}/docs/architecture/memory.md

@@ -46,7 +46,17 @@ A local-first system that gives LLMs persistent memory across conversations, wit
 
 ### Try It Now
 
-Get an LLM that remembers you using [Ollama](https://ollama.com). Two options:
+Get an LLM that remembers you using [Ollama](https://ollama.com). Three options:
+
+**Option 0: With Docker Compose (easiest)**
+
+```bash
+# Start Memory proxy with Ollama backend
+docker compose -f docker/docker-compose.yml --profile cpu up memory-proxy ollama
+
+# Point your chat client at http://localhost:8100/v1
+# The LLM will now remember facts across conversations
+```
 
 **Option A: With [Open WebUI](https://github.com/open-webui/open-webui) (web interface)**
 

{agent_cli-0.72.3 → agent_cli-0.72.5}/docs/architecture/rag.md

@@ -40,7 +40,19 @@ A local proxy that gives LLMs access to your documents using smarter multi-stage
 
 ### Try It Now
 
-Chat with your documents using [Ollama](https://ollama.com). Two options:
+Chat with your documents using [Ollama](https://ollama.com). Three options:
+
+**Option 0: With Docker Compose (easiest)**
+
+```bash
+# Start RAG proxy with Ollama backend
+docker compose -f docker/docker-compose.yml --profile cpu up rag-proxy ollama
+
+# Copy your documents into the volume
+docker cp ./my-docs/. $(docker volume inspect agent-cli-rag-docs --format '{{ .Mountpoint }}')/
+
+# Point your chat client at http://localhost:8000/v1
+```
 
 **Option A: With [Open WebUI](https://github.com/open-webui/open-webui) (web interface)**
 

{agent_cli-0.72.3 → agent_cli-0.72.5}/docs/installation/docker.md

@@ -59,7 +59,9 @@ The Docker setup provides:
 | ----------------------- | --------------------------------- | ----------- | ------------------------------ |
 | **whisper**             | agent-cli-whisper (custom)        | 10300/10301 | Speech-to-text (Faster Whisper)|
 | **tts**                 | agent-cli-tts (custom)            | 10200/10201 | Text-to-speech (Kokoro/Piper)  |
-| **transcribe-proxy** | agent-cli-transcribe-proxy     | 61337       | ASR proxy for iOS/external apps|
+| **transcribe-proxy**    | agent-cli-transcribe-proxy        | 61337       | ASR proxy for iOS/external apps|
+| **rag-proxy**           | agent-cli-rag-proxy               | 8000        | Document-aware chat (RAG)      |
+| **memory-proxy**        | agent-cli-memory-proxy            | 8100        | Long-term memory chat          |
 | **ollama**              | ollama/ollama                     | 11434       | LLM server                     |
 | **openwakeword**        | rhasspy/wyoming-openwakeword      | 10400       | Wake word detection            |
 
@@ -87,6 +89,19 @@ LLM_OLLAMA_MODEL=gemma3:4b  # Ollama model name
 LLM_OLLAMA_HOST=http://ollama:11434  # Ollama server URL (container name)
 LLM_OPENAI_MODEL=gpt-4.1-nano  # OpenAI model (if using openai provider)
 OPENAI_API_KEY=sk-...       # OpenAI API key (if using openai provider)
+
+# RAG Proxy
+RAG_PORT=8000               # Port for RAG proxy
+RAG_LIMIT=3                 # Number of document chunks per query
+RAG_ENABLE_TOOLS=true       # Enable read_full_document tool
+EMBEDDING_MODEL=text-embedding-3-small  # Embedding model for RAG/memory
+
+# Memory Proxy
+MEMORY_PORT=8100            # Port for memory proxy
+MEMORY_TOP_K=5              # Number of memories per query
+MEMORY_MAX_ENTRIES=500      # Max entries per conversation before eviction
+MEMORY_SUMMARIZATION=true   # Enable fact extraction from conversations
+MEMORY_GIT_VERSIONING=true  # Enable git versioning for memory changes
 ```
 
 ### GPU Support
@@ -117,11 +132,18 @@ Services store data in Docker volumes:
 - `agent-cli-tts-cache` - TTS models and voices
 - `agent-cli-ollama-data` - Ollama models
 - `agent-cli-openwakeword-data` - Wake word models
+- `agent-cli-rag-docs` - Documents to index for RAG
+- `agent-cli-rag-db` - RAG vector database (ChromaDB)
+- `agent-cli-rag-cache` - RAG embedding models
+- `agent-cli-memory-data` - Memory entries and vector index
+- `agent-cli-memory-cache` - Memory embedding models
 
 ## Ports Reference
 
 | Port  | Service             | Protocol |
 | ----- | ------------------- | -------- |
+| 8000  | RAG Proxy           | HTTP API |
+| 8100  | Memory Proxy        | HTTP API |
 | 10200 | TTS                 | Wyoming  |
 | 10201 | TTS                 | HTTP API |
 | 10300 | Whisper             | Wyoming  |