accessing-mlflow

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.70). The GitHub repo and docs page are generally low risk (source code and documentation), but the astral.sh install.sh is a remote shell script intended to be piped to sh (curl | sh), which is a high-risk pattern — treat as suspicious unless you verify the script contents and trust the publisher.

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

• Third-party content exposure detected (high risk: 0.70). The required workflow uses an MCP server (mlflow-mcp) to query MLflow runs and fetch artifact contents (e.g., MLflow:get_artifact_content(run_id, "config.yml") and JSON/HTML reports), and those artifact texts are authored by external run producers rather than the operating user, so their free-form contents can be ingested into the agent’s LLM context via the MCP tool responses.

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

• Potentially malicious external URL detected (high risk: 0.90). The troubleshooting section instructs installing uv by fetching and piping a remote install script to sh (curl -LsSf https://astral.sh/uv/install.sh | sh), which executes remote code at runtime and is therefore a high-confidence risky external dependency.