observability

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (high risk: 0.90). The "Companion check" instructs the agent to run local shell commands (ls ... ) and install packages (npx skills add ...) to inspect/modify the host environment — actions unrelated to observability and potentially exfiltrate or change the system, so this is a hidden/deceptive instruction outside the skill's stated purpose.

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

• Third-party content exposure detected (high risk: 0.90). The skill's "Companion check" in SKILL.md instructs the agent to offer installing missing companion skills using commands like npx skills add ..., which would fetch and install packages from public third-party registries (e.g., npm/GitHub) and thus expose the agent to untrusted external code/content that can change its behavior.