jeecg-aiflow

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill instructs the user to copy an X-Access-Token and show it in CLI commands (e.g., --token "") and store it in meta.json, which requires the agent/user to provide and embed the secret verbatim—an explicit high-risk secret-exfiltration pattern.

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

• Secret detected (high risk: 1.00). I inspected the entire skill codebase for literal, high-entropy credentials.

Findings:

• scripts/aiflow_apis.py contains a hardcoded signing secret:
• Variable: _SIGN_SECRET
• Value: "dd05f1c54d63749eda95f9fa6d49v442a"
• Why this is a secret: it's a non-trivial, fixed secret used to generate request signatures (see _get_sign). That gives it the ability to authenticate/forge signed API requests, so it is an actual credential and should be treated as sensitive (rotate and move to secure configuration).

Items I did NOT flag:

• Occurrences of placeholders or parameter names (e.g., "X-Access-Token", "--token", "", "<api_base>") — these are documentation/CLI placeholders and not literal credentials.
• Simple/default/example strings in templates (e.g., "start-node", "新流程", local API default "http://localhost:3100/jeecgboot") — low-entropy, not secret.
• Local filesystem paths and comments — not secrets.

Conclusion: one hardcoded secret present (the signing secret).