MCP Registry

@agent-infra/mcp-server-filesystem

f7672c63-db5f-5353-9147-d5fad004613d | Store v1.2.29
64/ 100
MEDIUM risk
AI review verdict
Review before use

The AI review rates the findings as likely false positive, but the risk score (64/100) still counts them.

Analysis record

Analysed
7 months ago 18 April 2026
Version
v1.2.29
Artifact
SHA256 20dc3588…c1e57c
Detail
Full list

Evidence ledger

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

45 findings

YARA Rule Matches

9 rules
SeverityRuleHitsFilesMetadata
HIGHpostinstall file download

File download activity detected

1
README.md
Risky Plugins Authors FP 30%
HIGHpostinstall obfuscation

Code obfuscation techniques detected

4
dist/index.cjsdist/server.cjsdist/index.js +1 more
Risky Plugins Authors FP 20%
HIGHcredential env files

Environment configuration file path detected

4
dist/index.cjsdist/server.cjsdist/index.js +1 more
Risky Plugins Authors FP 10%
HIGHNoUseWeakRandom

When software generates predictable values in a context requiring unpredictability, it may be possible for an attacker to guess the next value that will be generated, and use this guess to impersonate another user or access sensitive information. As the Math.random() function relies on a weak pseudorandom number generator, this function should not be used for security-critical applications or for protecting sensitive data. In such context, a cryptographically strong pseudorandom number generator (CSPRNG) should be used instead. For more information checkout the CWE-338 (https://cwe.mitre.org/data/definitions/338.html) advisory.

4
dist/index.cjsdist/index.jsdist/server.js +1 more
FP 5%
HIGHpostinstall system command

System command execution detected

8
dist/index.cjsdist/server.jsREADME.md +5 more
Risky Plugins Authors FP 10%
HIGHpostinstall network communication

Network communication detected

5
dist/index.cjsREADME.mddist/server.cjs +2 more
Risky Plugins Authors FP 30%
HIGHpostinstall file manipulation

File system manipulation detected

5
dist/index.cjsdist/server.cjsdist/index.js +2 more
Risky Plugins Authors FP 20%
HIGHpostinstall environment access

Environment variable access detected

1
dist/utils.d.ts
Risky Plugins Authors FP 40%
HIGHUsingShellInterpreterWhenExecutingOSCommands

Arbitrary OS command injection vulnerabilities are more likely when a shell is spawned rather than a new process, indeed shell meta-chars can be used (when parameters are user-controlled for instance) to inject OS commands. For more information checkout the CWE-78 (https://cwe.mitre.org/data/definitions/78.html) advisory.

2
dist/index.cjsdist/index.js
FP 10%

Finding Categories

34
Malware Signatures

YARA Rules Matched

9 rules(34 hits)
postinstall file download postinstall obfuscation credential env files NoUseWeakRandom postinstall system command postinstall network communication postinstall file manipulation postinstall environment access UsingShellInterpreterWhenExecutingOSCommands

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 zero tool-poisoning findings (confirmed in findings_summary: "tool-poisoning":"0"). This is the defining threat for MCP servers, and its absence is significant. There are no findings with titles like "tool-poisoning" or evidence of hidden AI manipulation directives embedded in tool descriptions.

Credential and Network Access: The evidence shows zero network findings ("network":"0") and zero secret findings ("secret":"0"). There are no credential-access findings targeting sensitive paths like .ssh/, .aws/credentials, .kube/config, or environment variables like GITHUB_TOKEN or AWS_SECRET_ACCESS_KEY. The package has no evidence of the exfiltration architecture (credential harvest + network to unknown domain) that defines real MCP threats.

IoC Finding Analysis: All 106 IoC findings are false positives from the XIOC extractor misreading JavaScript code as domains. Specific examples include:

  • XIOC-DOMAIN-objectprototypetostring.call - a property access chain, not a domain
  • XIOC-DOMAIN-o.constructor.name - JavaScript constructor access, not a domain
  • XIOC-DOMAIN-date.now - a method call, not a domain
  • XIOC-DOMAIN-program.name - variable access, not a domain
  • XIOC-DOMAIN-options.host - configuration property, not a domain

These match the documented IoC extractor garbage pattern: property access chains misread as domains. None are actual suspicious domains like evil.com or data-stealer.net.

Malware-Signature Findings: The 34 malware-signature findings are likely from YARA code-smell rules or bundled dependencies in dist/ files. With zero network activity and zero tool-poisoning, these signatures lack the behavioral context that would indicate real malware.

Strongest Counterargument: The package has 167 total findings and an anonymous publisher (oanakiaja) with zero users. However, finding COUNT is meaningless in MCP packages due to bundled dependencies and IoC noise. The NATURE of findings matters: zero tool-poisoning, zero network, zero secrets means no evidence of malicious behavior regardless of total count. Anonymous publishers are more suspicious in MCP, but without behavioral evidence of malice, this alone does not justify a malicious verdict.

Conclusion: This is a legitimate filesystem MCP server with false-positive findings from IoC extractor garbage and code-smell rules. The absence of tool-poisoning, network activity, and credential access findings confirms no malicious architecture.

Key Reasons

  • Zero tool-poisoning findings (defining MCP threat)
  • Zero network findings (no exfiltration capability)
  • Zero secret/credential findings (no credential theft)
  • All 106 IoC findings are property access chains, not real domains
  • Package serves legitimate filesystem access purpose

False Positive Considerations

  • IoC extractor misreading JavaScript property access chains as domains
  • YARA code-smell rules matching normal Node.js patterns
  • Bundled dependencies triggering multiplicative false positives
  • High finding count from IoC noise, not malicious behavior

Reviewed 2026-04-27; recommended action: suppress false positive; model confidence 85%.

Source Code Viewer

Source Code Not Available

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

About This Extension

MCP server for filesystem access

Frequently Asked Questions