linkedin-ads

SUSPICIOUS: The skill’s purpose and capabilities are broadly aligned, and the named API domain is same-brand. The main risk is install/execution trust: it asks the user to run an unpinned external MCP package and pass an API key into it, with only partial public provenance verification from the provided evidence.

SUSPICIOUS: the skill’s main advertised purpose is benign marketing guidance, but it unnecessarily pushes a specific third-party service, asks for an API key, and instructs execution of an unpinned external MCP package. Data flows stay on same-brand ColdIQ endpoints, so this is not confirmed malware, but the vendor-coupled install and credential forwarding are broader than needed for a campaign-setup skill.

SUSPICIOUS: the stated purpose is mostly coherent, and the ColdIQ API domain appears same-org and relevant to audience/ad research. But the skill expands into remote tool installation via unpinned `npx -y @coldiq/mcp@latest` and forwards an API key into that fetched package, creating medium supply-chain and credential-forwarding risk disproportionate to a simple copywriting guide.

SUSPICIOUS: the core guidance is benign marketing advice, but the skill’s footprint is broader than its stated LinkedIn bidding purpose because it pushes optional installation of external ColdIQ MCP tooling and credential use. Data flows stay within same-brand ColdIQ endpoints, so this is not clearly malicious, but the unpinned `npx` execution and third-party credential forwarding make it medium risk.

SUSPICIOUS: the skill’s main purpose is coherent, but it mixes a LinkedIn measurement guide with an optional external ColdIQ gateway and unpinned MCP install. The biggest risks are remote-code execution via `npx @latest`, forwarding an API key into a CLI package, and routing attribution data through a third-party marketplace layer rather than only direct platform integrations.

SUSPICIOUS: the skill’s marketing-audience purpose is coherent, but it pushes users toward an unpinned third-party MCP install and forwards an API key into externally fetched code whose exact package provenance was not verified. This is mainly a supply-chain and credential-forwarding risk, not confirmed malicious behavior.