Dockerfile.mcp-indexer

# Companion MCP server that exposes indexing/pruning/list tools over SSE
# Reuse the same deps as the indexer image so we can call the scripts directly.
FROM python:3.11-slim

ENV PYTHONDONTWRITEBYTECODE=1 \
    PYTHONUNBUFFERED=1 \
    WORK_ROOTS="/work,/app" \
    PIP_DEFAULT_TIMEOUT=120 \
    PIP_RETRIES=10 \
    LOG_LEVEL=INFO

# OS deps (git for history if we extend later, curl for model downloads, build-essential for compiling native extensions)
RUN apt-get update && apt-get install -y --no-install-recommends git ca-certificates curl build-essential \
    && rm -rf /var/lib/apt/lists/*

# Python deps: reuse shared requirements (includes FastMCP + OpenAI SDK)
COPY requirements.txt /tmp/requirements.txt
RUN pip install --no-cache-dir --upgrade --timeout=${PIP_DEFAULT_TIMEOUT} --retries=${PIP_RETRIES} -r /tmp/requirements.txt

# Create rerank dirs + download reranker model/tokenizer in single layer
ARG RERANKER_ONNX_URL=https://huggingface.co/cross-encoder/ms-marco-MiniLM-L-6-v2/resolve/main/onnx/model.onnx
ARG TOKENIZER_URL=https://huggingface.co/BAAI/bge-base-en-v1.5/resolve/main/tokenizer.json
RUN mkdir -p /tmp/rerank_events /tmp/rerank_weights \
    && chmod 777 /tmp/rerank_events /tmp/rerank_weights \
    && mkdir -p /app/models \
    && curl -L --fail --retry 3 -o /app/models/reranker.onnx "${RERANKER_ONNX_URL}" \
    && curl -L --fail --retry 3 -o /app/models/tokenizer.json "${TOKENIZER_URL}"

# Set default paths for reranker (can be overridden via env)
ENV RERANKER_ONNX_PATH=/app/models/reranker.onnx \
    RERANKER_TOKENIZER_PATH=/app/models/tokenizer.json

# Bake scripts into the image so entrypoints don't rely on /work
COPY scripts /app/scripts

COPY bench /app/bench

# Expose SSE port for this companion server
EXPOSE 8001

WORKDIR /work

# Default command runs the companion MCP server
CMD ["python", "/app/scripts/mcp_indexer_server.py"]