app-ai-guardrails

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt requires collecting a "SonarCloud org/key" as a parameter and explicitly demands echoing the full parameter block (including deferred/provided values), which forces the agent to output secret credential values verbatim.

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.70). The list contains a direct-install shell script URL (https://astral.sh/uv/install.sh), which is a high-risk distribution vector because downloadable/runnable install scripts can deliver arbitrary code and should be treated with caution; the other links are to known vendor sites, local host, or code repos and are lower risk in this context.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly fetches and installs third-party skills at runtime (e.g., a-tokyo/agent-skills, github/awesome-copilot, vercel-labs/agent-skills via npx skills add), which the runtime then loads as outsider-authored SKILL.md instructions, and it also documents runtime downloads that may be executed (https://astral.sh/uv/install.sh and https://start.spring.io/starter.tgz), so these fetched artifacts can directly control agent prompts or execute code.