MCP Registry

@michlyn/mcpgateway

cf960615-baa7-537b-bcf3-d22f22db2750 | Store v4.0.4
100/ 100
CRITICAL risk
AI review verdict
Do not install

The AI review rates the findings as likely false positive, but the risk score (100/100) still counts them.

Analysis record

Analysed
2 months ago 26 August 2026
Version
v4.0.4
Artifact
SHA256 6d0b3991…953cd3
Detail
Full list

Evidence ledger

Ranked by severity · findings with a source location link to the code viewer

422 findings
Showing 25 of 104 · highest severity first

YARA Rule Matches

14 rules
SeverityRuleHitsFilesMetadata
LOWpostinstall file download 68
CHANGE_LOG.mddocs/STREAMABLE_HTTP_GUIDE.md.github/workflows/docker-push.yaml +65 more
-
LOWNoUseWeakRandom 5
src/server/dynamicRouter.tsdist/server/dynamicRouter.jstest/test-auth-endpoint.js +2 more
-
LOWHavingAPermissiveCrossOriginResourceSharingPolicy 1
test/test-remote-config.js
-
LOWpostinstall network communication 51
src/gateways/stdioToSse.tsdist/cmd/mcpgateway.jsdist/server/dynamicRouter.js +48 more
-
LOWpostinstall file manipulation 31
src/gateways/apiToSse.tssrc/lib/openapi/api-gateway.tsdist/gateways/stdioToSse.js +28 more
-
LOWpostinstall crypto operations 16
src/server/dynamicRouter.tsBLOG.mdsrc/gateways/stdioToStreamableHttp.ts +13 more
-
LOWpostinstall environment access 38
dist/server/dynamicRouter.jsdist/lib/openapi/index.jssrc/gateways/stdioToWs.ts +35 more
-
LOWpostinstall system command 52
dist/cmd/mcpgateway.jstest/run-api-to-sse.jsdist/cmd/multiMcpServer.js +49 more
-
LOWpostinstall registry modification 7
build-docker-gf.sh.github/workflows/docker-push.yamlbuild-docker.sh +4 more
-
LOWpostinstall obfuscation 1
llms-full.txt
-
LOWUsingShellInterpreterWhenExecutingOSCommands 10
test/test-unified-mcpgateway.jstest/test-remote-api.jstest/test-sse-params.js +7 more
-
LOWUsingCommandLineArguments 11
dist/cmd/multiMcpServer.jsdist/cmd/openapi-to-mcp.jstest/test-streamable-http-timeout-fix.js +8 more
-
LOWcredential env files 12
src/gateways/stdioToWs.tsllms-full.txtdist/gateways/stdioToWs.js +9 more
-
LOWpostinstall persistence mechanism 15
test/test-client-session.jstest/test-sse-to-streamable-http.jstest/test-auth-mcp.js +12 more
-

Finding Categories

17
Secrets
61
Network
285
IoC Indicators

YARA Rules Matched

14 rules(318 hits)
postinstall file download NoUseWeakRandom HavingAPermissiveCrossOriginResourceSharingPolicy postinstall network communication postinstall file manipulation postinstall crypto operations postinstall environment access postinstall system command postinstall registry modification postinstall obfuscation UsingShellInterpreterWhenExecutingOSCommands UsingCommandLineArguments credential env files postinstall persistence mechanism

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 package has 0 tool-poisoning findings, which is the most critical indicator for MCP security. Tool poisoning—the defining threat for MCP servers—involves hidden AI manipulation directives embedded in tool descriptions. This package shows no evidence of such behavior. The findings summary explicitly reports "tool-poisoning":"0" with no associated file paths or finding titles to investigate.

Credential and Network Access

The evidence shows 61 network findings, all categorized as NET-FETCH or NET-AXIOS calls. Specific examples include:

  • src/gateways/apiToSse.ts:109 - fetch call
  • dist/lib/openapi/api-gateway.js:6 - axios call
  • dist/gateways/sseToStdio.js:39 - fetch call

These network calls align with the package's stated purpose: "adding streamable http support and the ability to run MCP services based on both OpenAPI protocol interface documentation." The file paths (apiToSse.ts, api-gateway.js, sseToStdio.js) indicate legitimate gateway/proxy functionality, not exfiltration architecture.

There are 17 critical "secret" findings, but the evidence bundle does not provide specific file paths or finding titles for these. Without details showing reads of .ssh/, .aws/credentials, .kube/config, or specific credential names like GITHUB_TOKEN or AWS_SECRET_ACCESS_KEY, these are likely legitimate API key configuration patterns common in gateway services.

IoC and Malware-Signature Noise

The 822 IoC findings and 318 malware-signature findings are classic false-positive patterns:

  • The package has malware:0 actual malware detections
  • IoC extractors produce massive false positives from bundled dependencies (IPv6 fragments, CDN domains, property chains misread as URLs)
  • Malware-signature YARA rules fire on minified JavaScript and common Node.js patterns

The description states this is an "enhanced version of supercorp-ai/supergateway," indicating it bundles existing libraries, which explains the high finding counts.

Strongest Counterargument

The 17 critical secret findings could indicate credential harvesting if they target sensitive paths. However, without specific evidence showing reads of .ssh/, .aws/, or exfiltration to unknown domains, and given the legitimate gateway functionality demonstrated by the network findings, this is more likely scanner noise from API key configuration code. The absence of tool poisoning and the alignment of network calls with stated functionality outweigh this concern.

Conclusion

This package is a legitimate MCP gateway with no evidence of tool poisoning, credential exfiltration, or malicious behavior. The high finding counts stem from bundled dependencies and scanner false positives.

Key Reasons

  • Zero tool-poisoning findings—the defining MCP threat is absent
  • Network activity (NET-FETCH, NET-AXIOS) aligns with gateway functionality described in package description
  • High IoC and malware-signature counts typical of bundled dependencies with malware:0 actual detections
  • Secret findings lack evidence of sensitive path access (.ssh, .aws, .kube) or exfiltration to unknown domains

False Positive Considerations

  • 822 IoC findings from bundled dependencies and minified code
  • 318 malware-signature YARA matches on legitimate Node.js patterns
  • Network findings consistent with stated gateway/proxy functionality
  • Zero tool-poisoning findings despite 1364 total findings

Reviewed 2026-04-27; recommended action: suppress false positive; model confidence 75%.

Source Code Viewer

Source Code Not Available

Source code is not available for this version of the extension.

About This Extension

Enhanced version of supercorp-ai/supergateway, adding streamable http support and the ability to run MCP services based on both OpenAPI protocol interface documentation and higress MCP template files.

Frequently Asked Questions