swanlab-skill

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly instructs supplying secrets via command-line flags (e.g., --api-key KEY and -p PASSWORD / create-user -p PASSWORD) and tells the agent to emit commands containing those values, which would force the LLM to include secret values verbatim in its output.

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

• Third-party content exposure detected (high risk: 0.90). The skill's CLI commands (e.g., swanlab api run logs, swanlab api run metrics, swanlab api run medias, and project/run listing) explicitly fetch user-generated experiment logs, metrics, and media from SwanLab hosts as described in SKILL.md and references/SWANLAB_CONCEPTS.md (including presigned media URLs and ECHARTS media), so the agent will read and act on untrusted third-party content.

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

• Attempt to modify system services in skill instructions detected (high risk: 0.70). The skill exposes self-hosted management commands that require root/admin privileges and explicitly includes a create-user subcommand (state-changing, privileged operations on the instance), so it encourages privileged modifications to the host instance even though it doesn't modify OS files directly.