xiaohongshu-mcp-openclaw

Security checks across malware telemetry and agentic risk

Overview

This skill does what it says, but review it before use because it logs into your Xiaohongshu account and may leave a network-accessible local service running.

Use this only on a trusted machine and network. Before scanning the login QR code, pin or review the upstream xiaohongshu-mcp version, make sure the MCP server is bound to 127.0.0.1 or firewalled, and avoid installing the keepalive service unless you really need it.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Risk analysis

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

#
ASI07: Insecure Inter-Agent Communication
Medium
What this means

On a shared or networked machine, another process or device that can reach the port might be able to interact with the MCP server using the logged-in Xiaohongshu session.

Why it was flagged

The script prints/checks a localhost endpoint but starts the MCP server with :PORT, which commonly binds all interfaces, and the launch command shows no authentication or host restriction.

Skill content
HOST="${XHS_MCP_HOST:-127.0.0.1}" ... nohup "$MCP_BIN" -headless="$HEADLESS" -port ":$PORT" ... echo "[INFO] MCP endpoint: http://$HOST:$PORT/mcp"
Recommendation

Bind the server explicitly to 127.0.0.1 if supported, firewall the port, avoid installing the keepalive service on shared networks, and add authentication if remote access is needed.

#
ASI04: Agentic Supply Chain Vulnerabilities
Medium
What this means

A future upstream change could alter the MCP server that handles logged-in Xiaohongshu requests.

Why it was flagged

The setup helper installs the upstream xiaohongshu-mcp binary at the moving latest version, without a pinned version or checksum.

Skill content
VERSION="${XHS_MCP_VERSION:-latest}" ... go install "${target}@${VERSION}"
Recommendation

Pin XHS_MCP_VERSION to a reviewed release and install only from the intended upstream source.

#
ASI03: Identity and Privilege Abuse
Medium
What this means

The skill will act with the Xiaohongshu account that scans the QR code and can access account-visible content.

Why it was flagged

The skill uses a QR login flow and reuses the resulting account session, with QR material saved locally for display.

Skill content
登录态通常可复用 ... 把二维码保存到本地(默认 `~/.openclaw/workspace/xhs-login-qrcode.png`)
Recommendation

Only scan the QR code if you trust the skill and upstream MCP server; keep the QR file private and revoke/logout the session when no longer needed.

#
ASI10: Rogue Agents
Low
What this means

The MCP server can keep running after the immediate task, continuing to hold local state and account session access.

Why it was flagged

The optional service installer creates a persistent user-level service that restarts the MCP server automatically.

Skill content
<key>RunAtLoad</key> ... <key>KeepAlive</key> ... ExecStart=${MCP_BIN} -headless=${HEADLESS} -port :${PORT} ... Restart=always
Recommendation

Install the service only if you need always-on behavior; use service_status.sh and service_uninstall.sh to monitor or remove it.

#
ASI05: Unexpected Code Execution
Low
What this means

If someone passes malicious custom arguments or environment values, the shell could execute unintended commands.

Why it was flagged

The service-management helper uses eval on constructed shell commands; defaults are fixed, but untrusted service names or environment values would be unsafe.

Skill content
SERVICE_NAME="${1:-xiaohongshu-mcp}" ... run_cmd() { ... eval "$@"; }
Recommendation

Do not pass untrusted service names or env overrides; maintainers should replace eval with direct command arrays.