@ubean/ai@0.7.13
AI large-model integration for the ubean framework: defineAgent + defineAgentTool + Vue runtime (thin orchestration over the Vercel AI SDK)
AI Security Review
scanned 5d ago · by lpm-firewall-aiThe documented fallback combines OPENAI_API_KEY with a DeepSeek endpoint by default, and agent execution passes the selected key to that provider. Documentation makes the behavior visible, but does not prevent the default cross-service credential routing.
Static reason
No blocking static signals were detected.; source fingerprint signature matched known malicious package; routed for review
Trigger
An agent run or stream uses the environment fallback when no provider is configured and OPENAI_API_KEY is the available key.
Impact
The OpenAI credential may be disclosed to a different AI provider, where it could be exposed or misused.
Mechanism
The fallback selects OPENAI_API_KEY while defaulting the provider base URL to api.deepseek.com; model creation passes that key to the provider client.
Attack narrative
When the environment fallback is used, the package accepts OPENAI_API_KEY but defaults requests to DeepSeek. An agent run or stream builds the model with that key, so the OpenAI credential can be sent to api.deepseek.com without an explicit choice to route it there.
AI rationale
The documented fallback combines OPENAI_API_KEY with a DeepSeek endpoint by default, and agent execution passes the selected key to that provider. Documentation makes the behavior visible, but does not prevent the default cross-service credential routing.
Evidence
dist/core-DllixiG1.jsREADME.md
Network endpoints1
api.deepseek.com
Decision evidence
public snapshotAI called this Malicious at 94.0% confidence as Malware with low false-positive risk.
Evidence for block
- The fallback provider uses OPENAI_API_KEY as its final key source and defaults to https://api.deepseek.com/v1 when no base URL is configured.
- The selected provider API key is passed to the OpenAI-compatible client.
- The credential is used when an agent run creates its model and calls generateText.
Evidence against
- The README documents the fallback key order and DeepSeek default endpoint.
- The package has no install lifecycle scripts; the model request occurs when the agent's run method is called.
Behavioral surface
EnvironmentVars
UrlStrings
NoLicense
Source & flagged code
6 flaggeddist/core-DllixiG1.jsView file
•matchType = malicious_source_fingerprint_signature
signature = eee1828c8d3bb801
signatureType = suspicious_hashes
sourceLabel = final_verdict:malicious
matchedPackage = @ubean/ai@0.7.12
matchedPath = dist/core-DllixiG1.js
matchedIdentity = npm:QHViZWFuL2Fp:0.7.12
similarity = 1.000
shingleOverlap = 4
summary = package final verdict is malicious
High
Known Malware Source Fingerprint Signature
Source fingerprint signature matches a known malicious package signature; route for source-aware review.
dist/core-DllixiG1.jsView on unpkg32function envProvider() {
const apiKey = process.env.DEEPSEEK_API_KEY || process.env.UBEAN_AI_API_KEY || process.env.OPENAI_API_KEY;
const baseURL = process.env.UBEAN_AI_API_BASE || "h
High
Ai Review Evidence
The fallback provider uses OPENAI_API_KEY as its final key source and defaults to https://api.deepseek.com/v1 when no base URL is configured.
dist/core-DllixiG1.jsView on unpkg · L3274async function createModel() {
const { ai, createOpenAICompatible } = await loadAiSdk();
const { provider, modelId } = resolveModel(config.model);
return {
ai,
model: createOpenAICompatible({
ba
Findings
6 High1 Medium3 Low
HighKnown Malware Source Fingerprint Signaturedist/core-DllixiG1.js
HighAi Review Evidencedist/core-DllixiG1.js
HighAi Review Evidencedist/core-DllixiG1.js
HighAi Review Evidencedist/core-DllixiG1.js
HighAi Review EvidenceREADME.md
HighAi Review Evidencepackage.json
MediumEnvironment Vars
LowScripts Present
LowUrl Strings
LowNo License
Affected versions and remediation
This report applies to @ubean/ai@0.7.13.
See all versions of @ubean/ai for other recorded verdicts.
- Avoid installing @ubean/ai@0.7.13. Remove it from direct dependencies and check your lockfile for transitive copies.
- Choose an independently verified alternative or release. This report does not establish that other versions are safe.
- If this version ran, investigate the affected machine and build environment. Rotate credentials it could access and rebuild from a trusted environment.
Evidence last updated: .
Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.