install

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill prompt instructs the agent to obtain an auth token via MCP tools and pass it directly as a command-line argument (--auth-token {TOKEN}) when executing the Subframe CLI, causing the secret to be exposed in the command output and logs.

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

• Potentially malicious external URL detected (high risk: 0.70). The skill instructs execution of npx @subframe/cli@latest init, which fetches and runs external code from npm without an immutable digest or explicit organization provenance. This constitutes a weak/unknown external operational dependency.