/security-scan - AIclude Security Vulnerability Scanner
Scan MCP Servers and AI Agent Skills for security vulnerabilities. Returns existing scan results instantly if available, or registers the target and triggers a new scan automatically.
Usage
# Search by name (recommended - leverages existing scan results)
/security-scan --name <package-name> [options]
# Scan a local directory directly
/security-scan <target-path> [options]
Parameters
--name: Name of the MCP server or Skill to search (npm package name, GitHub repo, etc.)target-path: Local path to scan (use instead of --name)--type: Target type (mcp-server | skill) - auto-detected if omitted--profile: Sandbox isolation profile (strict | standard | permissive)--format: Report output format (markdown | json)--engines: Scan engines to use (comma-separated)
Examples
# Search security scan results for an MCP server
/security-scan --name @anthropic/mcp-server-fetch
# Search scan results for a Skill
/security-scan --name my-awesome-skill --type skill
# Scan local source code
/security-scan ./my-mcp-server
What It Checks
- Prompt Injection: Hidden prompt injection patterns targeting LLMs
- Tool Poisoning: Malicious instructions embedded in tool descriptions
- Command Injection: Unvalidated input passed to exec/spawn calls
- Supply Chain: Known CVEs in dependencies and malicious packages (typosquatting)
- Malware: Backdoors, cryptominers, ransomware, data stealers, and obfuscated code
- Permission Abuse: Excessive filesystem, network, or process permissions
Scan Engines
7 engines run in parallel for comprehensive coverage:
| Engine |
Description |
| SAST |
Static code pattern matching against YAML rule sets |
| SCA |
Dependency CVE lookup via OSV.dev, SBOM generation |
| Tool Analyzer |
MCP tool definition analysis (poisoning, shadowing, rug-pull) |
| DAST |
Parameter fuzzing (SQL/Command/XSS injection) |
| Permission Checker |
Filesystem, network, and process permission analysis |
| Behavior Monitor |
Runtime behavior pattern detection |
| Malware Detector |
Signature scanning, entropy analysis, backdoor detection |
Output
The report includes:
- Risk Level Summary (CRITICAL / HIGH / MEDIUM / LOW / INFO)
- Vulnerability List (code location, description, severity)
- Risk Assessment (what risks are present and their impact)
- Remediation Recommendations (how to fix each vulnerability)
Web Dashboard
View all scan results at https://vs.aiclude.com
1---2name: aiclude-security-scan3description: Aiclude Security Scan4---5# /security-scan - AIclude Security Vulnerability Scanner67Scan MCP Servers and AI Agent Skills for security vulnerabilities. Returns existing scan results instantly if available, or registers the target and triggers a new scan automatically.89## Usage1011```12# Search by name (recommended - leverages existing scan results)13/security-scan --name <package-name> [options]1415# Scan a local directory directly16/security-scan <target-path> [options]17```1819## Parameters2021- `--name`: Name of the MCP server or Skill to search (npm package name, GitHub repo, etc.)22- `target-path`: Local path to scan (use instead of --name)23- `--type`: Target type (`mcp-server` | `skill`) - auto-detected if omitted24- `--profile`: Sandbox isolation profile (`strict` | `standard` | `permissive`)25- `--format`: Report output format (`markdown` | `json`)26- `--engines`: Scan engines to use (comma-separated)2728## Examples2930```31# Search security scan results for an MCP server32/security-scan --name @anthropic/mcp-server-fetch3334# Search scan results for a Skill35/security-scan --name my-awesome-skill --type skill3637# Scan local source code38/security-scan ./my-mcp-server39```4041## What It Checks4243- **Prompt Injection**: Hidden prompt injection patterns targeting LLMs44- **Tool Poisoning**: Malicious instructions embedded in tool descriptions45- **Command Injection**: Unvalidated input passed to exec/spawn calls46- **Supply Chain**: Known CVEs in dependencies and malicious packages (typosquatting)47- **Malware**: Backdoors, cryptominers, ransomware, data stealers, and obfuscated code48- **Permission Abuse**: Excessive filesystem, network, or process permissions4950## Scan Engines51527 engines run in parallel for comprehensive coverage:5354| Engine | Description |55|--------|-------------|56| SAST | Static code pattern matching against YAML rule sets |57| SCA | Dependency CVE lookup via OSV.dev, SBOM generation |58| Tool Analyzer | MCP tool definition analysis (poisoning, shadowing, rug-pull) |59| DAST | Parameter fuzzing (SQL/Command/XSS injection) |60| Permission Checker | Filesystem, network, and process permission analysis |61| Behavior Monitor | Runtime behavior pattern detection |62| Malware Detector | Signature scanning, entropy analysis, backdoor detection |6364## Output6566The report includes:671. **Risk Level Summary** (CRITICAL / HIGH / MEDIUM / LOW / INFO)682. **Vulnerability List** (code location, description, severity)693. **Risk Assessment** (what risks are present and their impact)704. **Remediation Recommendations** (how to fix each vulnerability)7172## Web Dashboard7374View all scan results at https://vs.aiclude.com
javimosch