composio

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt instructs using a literal CLI key in a command (composio login --key <cli_key> --no-wait), which requires the agent/LLM to accept and emit a secret value verbatim (high exfiltration risk).

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). The docs contain risky supply-chain instructions (curl | bash from a remote URL) and an explicit example that disables permission checks ("permissionMode: 'bypassPermissions'"), both of which enable abuse or privilege bypass if followed or copied into production.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill includes a runtime install command that fetches and executes remote code via "curl -fsSL https://composio.dev/install | bash", so the URL https://composio.dev/install is a high-risk runtime dependency.