koishi-plugin-smmcat-yeyechat@0.1.44
名为【夜夜酱】的聊天系统
AI Security Review
scanned 6d ago · by lpm-firewall-aiThe network and credential flows inspected serve the plugin’s chat, image recognition, and search features and use configured API tokens and base URLs. Image recognition requires a nonempty token; no install-time behavior or unrelated credential exfiltration was identified.
Decision evidence
public snapshot- The manifest identifies a Koishi chat plugin and points to lib/index.js as its entry point.
- Image recognition sends the supplied image and prompt to the configured vision API using the configured vision token.
- Image recognition is skipped unless a vision base URL, token, and model are configured.
- The chat reply engine sends chat messages to the configured API base URL with the configured authorization token.
- Web search sends its query and configured DeepSeek token to the configured official API base URL.
- The vision token defaults to empty, and image recognition is skipped without it.
- The vision API base URL can be changed or cleared in configuration; source does not show unrelated credential collection or transmission.
- The package manifest has no lifecycle scripts. No install-time network behavior was identified.
Source & flagged code
6 flaggedThe manifest identifies a Koishi chat plugin and points to lib/index.js as its entry point.
package.jsonView on unpkg · L2The chat reply engine sends chat messages to the configured API base URL with the configured authorization token.
lib/index.jsView on unpkg · L2809Findings
Affected versions and remediation
This report applies to koishi-plugin-smmcat-yeyechat@0.1.44.
See all versions of koishi-plugin-smmcat-yeyechat for other recorded verdicts.
- Review the evidence and your use of koishi-plugin-smmcat-yeyechat@0.1.44 before allowing it. Restrict the permissions described in this report.
- Choose an independently verified alternative or release. This report does not establish that other versions are safe.
Evidence last updated: .
Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.