Claude Code Setup in the Current openclaw

[buddingengineers12345]

Claude Code Installation in OpenClaw Docker

Overview

The following lines are proposed for addition to the Dockerfile to enable Claude Code installation in the existing OpenClaw Docker setup.

---

Questions / Discussion Points

• What would be the best place to keep this code content within the Dockerfile?

---

Proposed Dockerfile Snippet

# Runs as root (already active) — installs to /usr/local/bin/claude and
# cleans up all root-owned artefacts so nothing leaks into the image layers.
RUN curl -fsSL https://claude.ai/install.sh -o /tmp/install.sh \
    && bash /tmp/install.sh \
    && cp "$(readlink -f /root/.local/bin/claude)" /usr/local/bin/claude \
    && chmod 755 /usr/local/bin/claude \
    && test -x /usr/local/bin/claude \
    && rm -rf /tmp/install.sh \
              /root/.local \
              /root/.cache \
              /root/.npm \
              /root/.claude \
              /root/.claude.json

# Create Claude config dir under sandbox user for correct ownership.
USER sandbox
RUN mkdir -p /sandbox/.claude \
    && chmod 700 /sandbox/.claude \
    && echo '{}' > /sandbox/.claude.json \
    && chmod 600 /sandbox/.claude.json

---

Notes

• The RUN block above executes as root to perform the installation, then copies the binary to /usr/local/bin/claude to make it globally accessible.
• All root-owned artefacts (/root/.local, /root/.cache, /root/.npm, /root/.claude, /root/.claude.json) are cleaned up to avoid leaking into image layers.
• The subsequent USER sandbox block ensures the Claude config directory (/sandbox/.claude) is created with correct ownership and permissions under the sandbox user.