Skip to main content
Glama

CVESentry MCP (Lite)

License: MIT Python: 3.10+ MCP Standard: Compatible

Autonomous Local Vulnerability Scanner, Fast CVE Search Engine & Dependency Audit Plane for AI Coding Agents

CVESentry MCP gives AI coding agents (Claude Desktop, Cursor, Devin, Roo-Code) instant (<5ms) vulnerability intelligence without rate-limits or slow external API keys. It audits dependency manifests (requirements.txt, package.json), identifies vulnerable versions, and provides automated upgrade paths.


πŸš€ Features (Lite Edition)

  • Instant Local CVE Search (search_cve): Query vulnerability records in sub-millisecond local SQLite memory with zero network dependencies.

  • Dependency Manifest Auditor (audit_manifest): Parses requirements.txt and package.json, checks versions against vulnerable ranges, and calculates severity breakdowns.

  • CVE Detailed Metadata (get_cve_details): Detailed CVSS score, vulnerability category, affected package version bounds, and security summaries.

  • Safe Version Recommendation (get_safe_recommendation): Calculates the minimal patch version that eliminates all matching CVEs.


Related MCP server: Security Intel MCP

πŸ“¦ Installation & Setup

git clone https://github.com/johannschiffer/cvesentry-mcp.git
cd cvesentry-mcp
pip install -e .

Configuration with Claude Desktop / Cursor

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "cvesentry": {
      "command": "python3",
      "args": ["-m", "cvesentry.server"]
    }
  }
}

⚑ Quick CLI Usage

# Search for a CVE or package
cvesentry search --query "aiohttp"

# Audit a requirements.txt file
cvesentry audit --file requirements.txt

πŸ’Ό Upgrade to CVESentry Pro

Need enterprise security compliance for automated code generation?

  • Multi-Ecosystem Engine (SQLite WAL): Full parsing for Python (pyproject.toml, Pipfile), Node (package-lock.json), Rust (Cargo.lock), and Go (go.mod).

  • EPSS & CISA KEV Exploit Prioritization: Real-world Exploit Prediction Scoring System (EPSS) and CISA Known Exploited Vulnerabilities catalog integration.

  • Automated Safe Version Upgrader: Automatically patches manifests while preserving semver constraints to minimize breaking changes.

  • Audit Certificate Exporter: Generates formatted markdown and JSON reports for GitHub PR reviews.

  • Docker & Compose Ready: Deploy as a microservice sidecar in your CI/CD cluster.

Available at Gumroad.


πŸ“„ License

MIT Β© Johann Schiffer

Available Tools

4 tools
audit_manifestA

Audits raw contents of requirements.txt or package.json for known vulnerabilities and returns fix recommendations.

Inputs: content (required), manifest_type (required).

get_cve_detailsA

Retrieves complete vulnerability metadata, CVSS score, severity, and affected ranges for a specific CVE ID.

Inputs: cve_id (required).

get_safe_recommendationB

Returns the minimum safe version required to patch vulnerabilities for a given package and version.

Inputs: package (required), ecosystem (required), current_version (required).

search_cveA

Searches local CVE database by ID (e.g., CVE-2024-23334), package name, or keyword with zero latency.

Inputs: query (required).

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A3.7/5.0

    Scored across 4 tools

    Disambiguation4/5

    Each tool has a distinct primary purpose, but there is minor overlap: search_cve can search by CVE ID while get_cve_details retrieves full details for a CVE ID, and audit_manifest returns fix recommendations that overlap with get_safe_recommendation's single-package fix output. The descriptions still provide enough guidance to select correctly.

    Naming Consistency5/5

    All four tool names follow a consistent snake_case verb_noun pattern (search_cve, audit_manifest, get_cve_details, get_safe_recommendation). There is no mixing of conventions or verb styles.

    Tool Count5/5

    Four tools are well-scoped for a local CVE lookup and remediation server: search, audit, detailed lookup, and safe version recommendation. Each tool earns its place and the set is not bloated or thin.

    Completeness4/5

    The set covers the core lifecycle: lookup by ID/package/keyword, audit a manifest, retrieve full CVE metadata, and get a safe patch version. A minor gap is the lack of a direct 'is this version vulnerable?' query, though agents can infer this from search_cve or get_safe_recommendation.

    Related MCP Connectors

    Related MCP Servers