visual-edit

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt requires capturing a server-minted "bridgeToken" and inserting it into shell commands or CLI flags (e.g., AGENT_NATIVE_BRIDGE_TOKEN="<bridgeToken from step 2>" or --bridge-token ), which forces the agent to handle and potentially output the secret verbatim.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). The document intentionally describes a token-authenticated local "bridge" that the remote Design service (via the browser) can use to call /read-file and /write-file on 127.0.0.1, creating a deliberate, high-risk remote access/exfiltration capability that could be abused to read or modify local files.