agent-tools

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.70). These URLs all point to a single third‑party domain that provides a remote install script (curl | sh) and direct binary downloads (manifest/checksums), which is potentially risky because executing remote shell scripts or unsigned binaries from a non‑well‑known source can install malware—although the presence of checksums and optional Sigstore verification mitigates but does not remove the risk.

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

• Third-party content exposure detected (high risk: 0.80). The skill's SKILL.md and references explicitly include web-search apps (e.g., infsh app run tavily/search-assistant and Exa Search listed in SKILL.md and references) which fetch and return open/public web content that the agent would read and could influence subsequent actions.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill explicitly tells users to run a remote installer with "curl -fsSL https://cli.inference.sh | sh" (which downloads binaries from https://dist.inference.sh), so remote code is fetched and executed as a required dependency.