vm-agentic-security-scanner

# vm-agentic-security-scanner

## When To Use This Skill

- When auditing an agent definition (.agent.md) for agentic security vulnerabilities
- When auditing a skill definition (SKILL.md) for agentic security vulnerabilities
- When checking whether an agent's tool set follows least privilege
- When verifying subagent trust boundary checks are present
- When detecting undeclared script execution in skill instructions

**Do NOT use this skill when:**
- Scanning application source code for traditional OWASP vulnerabilities (use `vm-vulnerability-scanner`)
- Scanning dependencies for CVEs (use `vm-dependency-scanner`)
- Performing language-specific code pattern review (use `vm-secure-coding-reviewer`)
- Building or modifying agent definitions (use `0 - Meta - AI Agent Builder`)

## Unitary Function

**ONE RESPONSIBILITY:** Evaluate agent and skill definition files against the 10-category agentic security threat taxonomy and produce structured findings.

**NOT RESPONSIBLE FOR:** Fixing agent definitions, scanning application source code, running external security tools, creating GitHub issues, or making REUSE/ADAPT/CREATE decisions about agents.

## Input

```json
{
  "file_path": "path to .agent.md or SKILL.md file",
  "file_content": "full text content of the file",
  "file_type": "agent | skill",
  "standards_baseline": {
    "section_10": "Security Standards text",
    "section_6_2": "Subagent Independence text",
    "section_3_5": "Tool Selection text",
    "section_3_3": "Authority & Boundaries text",
    "section_12_1": "Context Window Discipline text"
  }
}
```

| Parameter | Type | Required | Description |
|---|---|---|---|
| file_path | string | MUST | Absolute or relative path to the definition file |
| file_content | string | MUST | Full text content of the file |
| file_type | string | MUST | `agent` or `skill` |
| standards_baseline | object | MUST | Relevant sections from AGENT_DEVELOPMENT_STANDARDS.md |

## Output

```json
{
  "generated_by": {
    "skill": "vm-agentic-security-scanner",
    "version": "1.0.0"
  },
  "file_path": "path/to/file",
  "file_type": "agent | skill",
  "scan_status": "complete | partial | failed",
  "findings": [
    {
      "id": "ASA-NNN-001",
      "threat_category": "ASA-01",
      "threat_name": "Prompt Injection Defense",
      "severity": "Critical | High | Medium | Low",
      "evidence": "Quoted text from the file or structural observation",
      "location": "Section name or line range in the file",
      "remediation": "Specific action to fix this finding",
      "standards_reference": "Section 10.1 of AGENT_DEVELOPMENT_STANDARDS.md"
    }
  ],
  "scripts_detected": [
    {
      "script_reference": "Quoted instruction text mentioning script",
      "script_path": "path/to/script.py or null if undeclared",
      "language": "python | powershell | posix | unknown",
      "declared": true
    }
  ],
  "summary": {
    "total_findings": 0,
    "critical": 0,
    "high": 0,
    "medium": 0,
    "low