@mseep/hypefury-mcp-smithery
The AI review rates the findings as likely false positive, but the risk score (81/100) still counts them.
Analysis record
- Analysed
- 2 months ago 26 August 2026
- Version
- v1.0.2
- Artifact
- SHA256 88d0699b…ecde51
- Detail
- Full list
Evidence ledger
Ranked by severity · findings with a source location link to the code viewer
YARA Rule Matches
6 rules| Severity | Rule | Hits | Files | Metadata |
|---|---|---|---|---|
| LOW | credential env files | 2 | README.mdsrc/utils.ts | - |
| LOW | postinstall persistence mechanism | 1 | src/index.ts | - |
| LOW | postinstall file download | 2 | package.jsonsrc/utils.ts | - |
| LOW | postinstall environment access | 2 | src/utils.tssmithery.yaml | - |
| LOW | postinstall network communication | 3 | src/index.tsREADME.mdpackage.json | - |
| LOW | postinstall system command | 2 | README.mdDockerfile | - |
Finding Categories
YARA Rules Matched
6 rules(12 hits)MCP Server Analysis
MCP servers expose tools and resources to AI assistants. Unlike browser extensions, they run as standalone processes with direct system access. Tool definitions are analyzed for prompt injection, data exfiltration, and tool poisoning patterns.
AI Security Report
AI Security Review
Evidence context: threat category none; evidence quality moderate.
Tool Poisoning Assessment
The single finding titled MCP-TOOL-TOOL-POISONING-src/index.ts-72 is classified as code-smell with low severity, not as an actual tool-poisoning finding. The findings summary confirms "tool-poisoning":"0" with zero genuine tool-poisoning detections. According to CVEQ's known false-positive patterns, code-smell findings are noise and should not drive verdicts. This finding likely represents a legitimate tool definition in src/index.ts where the MCP server registers its tools with descriptions, which the scanner misidentifies as poisoning. There is no evidence of hidden AI manipulation directives, invisible Unicode characters, or XML-style instruction tags.
Credential and Network Access
The package contains one secret finding, but the evidence does not show this targets sensitive paths like .ssh/, .aws/credentials, .kube/config, or specific secret names (e.g., GITHUB_TOKEN, AWS_SECRET_ACCESS_KEY). The network findings show two NET-FETCH calls in src/utils.ts:36 and src/utils.ts:44. The detected domains are all legitimate and expected for this package:
github.com- standard code hostingsmithery.ai- Smithery is an MCP platform where this server is likely distributedapp.hypefury.com- the Hypefury service this MCP server connects tohttps://github.com/eugenechabanov/hf-mcp.git- GitHub repository URL
The combination of credential reads plus network calls to unknown domains is the exfiltration signal. Here, the network destinations are documented and legitimate (Hypefury is the service this MCP integrates with). There is no evidence of exfiltration architecture.
Malware Signatures
The 12 malware-signature findings are consistent with bundled dependencies in the dist/ directory, which is normal for MCP packages. The IoC count (5 findings) is meaningless in MCP packages where everything is bundled. The domains detected are all legitimate infrastructure, not suspicious external endpoints.
Strongest Counterargument
The 12 malware-signature findings could suggest malicious code. However, these are well-documented false positives from bundled npm packages in distribution files. The scanner's code-smell rules fire on almost any non-trivial JavaScript, and the findings summary shows zero actual tool-poisoning or credential-theft findings. The package's purpose (Hypefury integration) aligns with the network destinations observed.
Key Reasons
- Zero tool-poisoning findings in findings summary despite code-smell noise
- All network domains are legitimate (github.com, smithery.ai, app.hypefury.com)
- No credential access to sensitive paths (.ssh, .aws, .kube) in evidence
- Malware-signature findings consistent with bundled dependencies
- Package purpose (Hypefury MCP) matches observed network behavior
False Positive Considerations
- code-smell finding misidentified as tool-poisoning
- malware-signature hits from bundled dependencies
- IoC extraction noise from legitimate domains
- Network calls to documented service endpoints
Reviewed 2026-04-27; recommended action: suppress false positive; model confidence 75%.
Source Code Not Available
Source code is not available for this version of the extension.
About This Extension
Frequently Asked Questions
Similar Extensions
Related extensions from the same publisher or marketplace