xmake-remote-compilation

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The prompt explicitly shows and describes copying/embedding server tokens and passwords into client/server config files (including literal token examples) which encourages the agent to accept and output secret values verbatim into configs/commands, creating an exfiltration risk.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill configuration includes runtime connections such as connect = "192.168.56.110:9691" (and other host entries like 127.0.0.1:9691, 10.5.138.247:9691, 10.5.138.100:9691) which are used at runtime to execute every xmake command on the remote server (remote code execution).

HIGH W008: Secret detected in skill content (API keys, tokens, passwords).

• Secret detected (high risk: 1.00). The server/client configs include a literal, high-entropy token string "e438d816c95958667747c318f1532c0f" present in both ~/.xmake/service/server.conf and ~/.xmake/service/client.conf. This is a direct, likely-usable access token (not a placeholder or truncated example), so it meets the definition of a secret.

Other entries that look like tokens in the multi-host snippet are truncated with "..." (e.g., "ab9dcb6f...", "0e052f8c...", "7889e254...") — these are redacted/truncated and are ignored per the rules. IP:port values, config keys, and the absence of any plaintext password values or private key blocks mean no other secrets are present.