etienne

# Claude Multi-Tenant

## Components

* /frontend contains the React/Vite/Typescript browser frontend running on :5000
* /backend contains the nest.js/node.js 20 API backend running on :6060
* a docker image name claude-code contains the latest Claude Code developer image

The goal of the the project is to manage isolated projects in a central mounted volume /workspace.

## Embeddings

All embedding generation MUST go through `EmbeddingsService` in [backend/src/embeddings](backend/src/embeddings). Do not import `openai`, `@huggingface/transformers`, or any other embedding library directly from feature modules — inject `EmbeddingsService` and call `embed(text)` / `embedBatch(texts)`.

Default provider is the local self-hosted `transformers` (`Xenova/multilingual-e5-base`, ~768-dim, no API key required). OpenAI is opt-in via `EMBEDDING_PROVIDER=openai` (requires `OPENAI_API_KEY`). ChromaDB collections are dimension-namespaced (`documents_${dimension}`), so switching providers starts with empty collections — documents must be re-ingested.

## Project Structure

Each project in the workspace follows this structure:
```
workspace/<project-name>/
├── .claude/
│   ├── CLAUDE.md          # System prompt/role definition for Claude Code
│   └── settings.json      # Project-specific Claude Code settings
├── data/
│   └── permissions.json   # Tool permissions
└── out/                   # Output files
```

## Working with PDF and Office Format Documents

The standard skill **office-and-pdf-documents** provides document parsing capabilities using the `@llamaindex/liteparse` npm package. It extracts text from documents while preserving spatial layout, and includes built-in OCR for scanned content. LiteParse is the local, open-source version of LlamaIndex's cloud service **LlamaParse**, which could be used for use cases with more complex requirements in regard to parsing quality (e.g. complex tables, multi-column layouts, or heavily formatted documents).

### Supported formats

- **PDF** — native digital PDFs as well as PDFs containing scanned images (processed via built-in OCR)
- **Word** — `.doc`, `.docx`, `.docm`, `.odt`, `.rtf`
- **PowerPoint** — `.ppt`, `.pptx`, `.pptm`, `.odp`
- **Excel / Sheets** — `.xls`, `.xlsx`, `.xlsm`, `.ods`, `.csv`, `.tsv`
- **Images** — `.jpg`, `.jpeg`, `.png`, `.gif`, `.bmp`, `.tiff`, `.webp`, `.svg` (via OCR)

### Binary dependency: LibreOffice

Non-PDF formats (Word, PowerPoint, Excel) are converted to PDF before parsing. This conversion requires **LibreOffice** (`soffice`) to be installed on the system. It is a binary dependency that cannot be installed via npm and must be installed manually.

Before parsing, the skill checks availability by running:

```bash
soffice --version
```

If LibreOffice is missing and the user needs to parse Office documents, the skill will report the issue and provide installation instructions. PDF parsing works without LibreOffice.

### How to start claude code container (obsolete)
```bash
docker run -it \
  --name claude-code \
  -v C:/Data/GitHub/claude-multitenant/workspace:/workspace \
  -e ANTHROPIC_API_KEY=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 \
  claude-code
```

### claude-code docker file (obsolete)
```
FROM node:20

ARG TZ
ENV TZ="$TZ"

ARG CLAUDE_CODE_VERSION=latest

# Install basic development tools and iptables/ipset
RUN apt-get update && apt-get install -y --no-install-recommends \
  less \
  git \
  procps \
  sudo \
  fzf \
  zsh \
  man-db \
  unzip \
  gnupg2 \
  python3 \
  python3-pip \
  gh \
  iptables \
  ipset \
  iproute2 \
  dnsutils \
  aggregate \
  jq \
  nano \
  vim \
  && apt-get clean && rm -rf /var/lib/apt/lists/*

# Ensure default node user has access to /usr/local/share
RUN mkdir -p /usr/local/share/npm-global && \
  chown -R node:node /usr/local/share

ARG USERNAME=node

# Persist bash history.
RUN SNIPPET="export PROMPT_COMMAND='history -a' && export HISTFILE=/commandhistory/.bash_history" \
  && mkdir /commandhistory \
  && touch /commandhistory/.bash_history \
  && chown -R $USERNAME /commandhistory

# Set `DEVCONTAINER` environment variable to help with orientation
ENV DEVCONTAINER=true

# Create workspace and config directories and set permissions
RUN mkdir -p /workspace /home/node/.claude && \
  chown -R node:node /workspace /home/node/.claude

WORKDIR /workspace

ARG GIT_DELTA_VERSION=0.18.2
RUN ARCH=$(dpkg --print-architecture) && \
  wget "https://github.com/dandavison/delta/releases/download/${GIT_DELTA_VERSION}/git-delta_${GIT_DELTA_VERSION}_${ARCH}.deb" && \
  sudo dpkg -i "git-delta_${GIT_DELTA_VERSION}_${ARCH}.deb" && \
  rm "git-delta_${GIT_DELTA_VERSION}_${ARCH}.deb"

# Switch to root to modify sudoers
USER root

# Allow node user to install Python packages
RUN echo 'node ALL=(ALL) NOPASSWD: /usr/bin/pip, /usr/bin/pip3, /usr/local/bin/pip, /usr/local/bin/pip3' >> /etc/sudoers

# Set up non-root user
USER node

# Install global packages
ENV NPM_CONFIG_PREFIX=/usr/local/share/npm-global
ENV PATH=$PATH:/usr/local/share/npm-global/bin

# Set the default shell to zsh rather than sh
ENV SHELL=/bin/zsh

# Set the default editor and visual
ENV EDITOR=nano
ENV VISUAL=nano

# Default powerline10k theme
ARG ZSH_IN_DOCKER_VERSION=1.2.0
RUN sh -c "$(wget -O- https://github.com/deluan/zsh-in-docker/releases/download/v${ZSH_IN_DOCKER_VERSION}/zsh-in-docker.sh)" -- \
  -p git \
  -p fzf \
  -a "source /usr/share/doc/fzf/examples/key-bindings.zsh" \
  -a "source /usr/share/doc/fzf/examples/completion.zsh" \
  -a "export PROMPT_COMMAND='history -a' && export HISTFILE=/commandhistory/.bash_history" \
  -x

# Install Claude
RUN npm install -g @anthropic-ai/claude-code@${CLAUDE_CODE_VERSION}


# Copy and set up firewall script
COPY init-firewall.sh /usr/local/bin/
USER root
RUN chmod +x /usr/local/bin/init-firewall.sh && \
  echo "node ALL=(root) NOPASSWD: /usr/local/bin/init-firewall.sh" > /etc/sudoers.d/node-firewall && \
  chmod 0440 /etc/sudoers.d/node-firewall
USER node
```