ln-012-mcp-configurator

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.90). The skill explicitly calls external services and tools (e.g., npm view, HTTP MCP endpoints like https://api.ref.tools and https://mcp.context7.com, and—most critically—calls the hex-graph-mcp tool and "reuse[s] its instructions_for_agent output as the remediation text" in Phase 2b Step 4 of SKILL.md), causing the agent to ingest and act on arbitrary third-party responses that can materially influence subsequent actions.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly runs remote packages at runtime (e.g., "npx -y @levnikolaevich/hex-line-mcp", which fetches and executes code from the npm registry) and registers required HTTP MCP endpoints such as https://api.ref.tools/mcp (and https://mcp.context7.com/mcp / https://mcp.linear.app/mcp), so external content is fetched/used at runtime and can execute code or directly control agent behavior.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 0.90). The skill instructs the agent to install system packages (including explicit "sudo apt-get" commands), run global tool installs, and write/modify user and IDE configuration files (and could require elevated privileges), which are state-changing operations that can compromise the host if misused even though some actions require user consent.