registry / koishi-plugin-smmcat-yeyechat / 0.1.45
koishi-plugin-smmcat-yeyechat@0.1.45
名为【夜夜酱】的聊天系统
AI Security Review
scanned 6d ago · by lpm-firewall-aiThe inspected network and file operations are tied to the plugin’s chatbot, image, search, and voice features. Tokens are used with their corresponding configured API bases, and there is no install-time behavior or unrelated data collection in the inspected source.
Static reason
No blocking static signals were detected.
Trigger
Runtime chat activity, incoming images, or a model-selected search/fetch or voice feature; no install hook is declared.
Impact
Chat content or images may be sent to the configured AI service for the requested feature; no unrelated collection or account action was found.
Mechanism
Uses configured API bases and their corresponding tokens for feature requests. Image recognition is disabled without its configuration token and other required settings.
AI rationale
The inspected network and file operations are tied to the plugin’s chatbot, image, search, and voice features. Tokens are used with their corresponding configured API bases, and there is no install-time behavior or unrelated data collection in the inspected source.
Evidence
lib/index.jsctx.baseDir + imageSavePath (default /data/smmAiImg)
Network endpoints7
wxls.cc.cd/v1/chat/completionsapi.deepseek.com/tools/searchapi.deepseek.com/tools/fetchcp.compshare.cn182.92.130.139:3060/generateConfigured config.baseUrl API endpointUser-provided image URLs and fetch_url targets
Decision evidence
public snapshotAI called this Clean at 90.0% confidence as Benign with low false-positive risk.
Evidence for AI clean decision
- The manifest identifies a Koishi chatbot plugin, points to lib/index.js, and contains no lifecycle scripts.
- Image recognition sends an Authorization token to the configured vision API; recognition is skipped unless its base URL, token, and model are set.
- The vision API defaults to https://wxls.cc.cd, and its token is empty by default; the setting is described as a Gemini-compatible vision interface.
- Incoming images trigger image saving and the optional vision-description flow, consistent with the chatbot’s stated purpose.
- The DeepSeek token is used for the search tool request at the configured official API base.
Evidence against
Behavioral surface
CryptoFilesystemNetwork
HighEntropyStringsUrlStrings
Source & flagged code
6 flaggedpackage.jsonView file
1{
"name": "koishi-plugin-smmcat-yeyechat",
"description": "名为【夜夜酱】的聊天系统",
"version": "0.1.45",
"main": "lib/index.js",
"typings": "lib/index.d.ts",
"files": [
"lib",
"dist"
],
"license": "MIT",
"keywords": [
"chatbot",
"koishi",
"plugin"
],
"peerDependencies": {
"koishi": "^4.18.7"
}
}
Low
Suspicious Lifecycle Evidence
The manifest identifies a Koishi chatbot plugin, points to lib/index.js, and contains no lifecycle scripts.
package.jsonView on unpkg · L1lib/index.jsView file
85async function callGemini(ctx, config, imageUrl) {
const authHeader3 = config.visionToken.startsWith("Bearer ") ? config.visionToken : `Bearer ${config.visionToken}`;
const e
Medium
Stripped Provenance Metadata
Image recognition sends an Authorization token to the configured vision API; recognition is skipped unless its base URL, token, and model are set.
lib/index.jsView on unpkg · L85Findings
3 Medium7 Low
MediumNetwork
MediumStripped Provenance Metadatalib/index.js
MediumStripped Provenance Metadatalib/index.js
LowFilesystem
LowHigh Entropy Strings
LowUrl Strings
LowSuspicious Lifecycle Evidencepackage.json
LowAi Review Evidencelib/index.js
LowAi Review Evidencelib/index.js
LowAi Review Evidencelib/index.js
Affected versions and remediation
This report applies to koishi-plugin-smmcat-yeyechat@0.1.45.
See all versions of koishi-plugin-smmcat-yeyechat for other recorded verdicts.
- Review the evidence and your use of koishi-plugin-smmcat-yeyechat@0.1.45 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.