deployment

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly accepts HuggingFace model IDs and inspects public checkpoint files (e.g., SKILL.md and scripts/deploy.sh: detect_quantization reads hf_quant_config.json or infers from HF model names, and references pulling models from the HuggingFace hub and public container registries), so it ingests untrusted, user-provided third‑party artifacts whose contents (names/configs) directly influence deployment commands and flags.