MCP Registry

@argosvix/mcp-server

59a3138f-fbf1-5c59-934c-4f44d86263b2 | Store v1.1.2
100/ 100
CRITICAL risk
No change since v1.1.1
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. The review covers version 0.2.0-alpha.1, not the analysed version 1.1.2.

Analysis record

Analysed
2 months ago 3 September 2026
Version
v1.1.2
Artifact
SHA256 ecfeb688…d109d3
Detail
Full list

Evidence ledger

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

89 findings
Showing 25 of 89 · highest severity first

Finding Categories

84
Secrets
5
Network

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.

This MCP server shows no evidence of the defining threats for this package type. There are zero tool-poisoning findings, which is the most critical indicator for MCP servers. The 30 IoC findings all reference legitimate, documented domains: argosvix.com and ingest.argosvix.com are the package's own service endpoints matching the package name @argosvix/mcp-server, while img.shields.io and github.com are public infrastructure. The finding XIOC-URL-https://dashboard.argosvix.com/api-keys references the package's documented API key management interface, not exfiltration.

The 20 critical-severity secret findings are code-smell detections matching generic credential patterns like apiKey or password in configuration objects. These are standard for any MCP server requiring authentication to query the Argosvix API. There is no evidence of credential harvesting targeting sensitive paths like .ssh/, .aws/credentials, or .kube/config. The single network finding corresponds to expected API calls to the documented Argosvix endpoints.

The strongest counterargument is the high critical-severity count (20 findings). However, these are credential-access code-smell rules that fire on any code referencing API keys or environment variables—every legitimate MCP server triggers these. Without evidence of the credential-exfiltration architecture (credential reads combined with network calls to unknown domains), these findings represent normal configuration patterns. The package's description explicitly states it queries Argosvix traces/cost/alerts, which requires API credentials.

The zero tool-poisoning findings, combined with all network destinations matching documented package domains and no suspicious credential access patterns, indicates this is a legitimate MCP server with standard security scanner noise.

Key Reasons

  • Zero tool-poisoning findings despite being the defining MCP threat
  • All IoC domains match documented package services (argosvix.com, ingest.argosvix.com)
  • Secret findings are generic credential-access patterns, not targeted theft
  • No network calls to unknown or suspicious domains

False Positive Considerations

  • Generic credential-access code-smell rules matching API key configuration
  • IoC extraction from legitimate service URLs in documentation
  • Standard MCP server authentication patterns flagged as critical

Reviewed 2026-06-01 · version 0.2.0-alpha.1; recommended action: suppress false positive; model confidence 85%.

MCP version history

Risk trend by version

19 analyzed versions. Each point is the latest successful scan for that version; failed zero-score scans are hidden. Dates: First seen by Risky Plugins.

Selected
100
Change since first
No change
Change from previous
No change
Versions:
First analyzed version
0.2.0-alpha.1
First seen Jun 1, 2026
Risk range
100 to 100
Across analyzed versions
Latest analyzed version
1.1.2
First seen Sep 3, 2026
Selected version
critical
Version
v1.1.2
First seen 2 months ago
Risk score
100
Findings
89
Change vs previous
No change

Pick any point on the chart, then open the Source tab to read that version's code.

Source Code Viewer

Source Code Not Available

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

About This Extension

Observability MCP server for AI agents: 89 tools (2 usable with no API key) to query and operate LLM cost, errors, latency, alerts, budgets, evals, prompts, approvals, and safety checks from Claude Desktop, Cursor, or any MCP client. stdio + HTTP transpor

Frequently Asked Questions