agent-tools

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.70). The URLs point to a third‑party service that publishes an executable installer (curl | sh via https://cli.inference.sh) and binary distribution endpoints (dist.inference.sh), which is not a major trusted vendor — the presence of checksums and optional Sigstore verification reduces but does not eliminate the risk of executing untrusted binaries, so this is potentially unsafe to run without independent verification.

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

• Third-party content exposure detected (high risk: 0.80). The SKILL.md explicitly shows running web-search apps (e.g., "infsh app run tavily/search-assistant" in Quick Examples and the "Search" entries under "What's Available"), which indicates the agent will fetch and ingest untrusted public web/search results as part of its workflow and could act on that content.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill instructs installing the required CLI with a direct piped installer (curl -fsSL https://cli.inference.sh | sh), which fetches and executes remote code as a required dependency for running the skill.