Copyright (c) 2026 Veritas Aequitas Holdings LLC. All rights reserved.

Malwar GitHub Action

Scan SKILL.md files in pull requests for malware, prompt injection, data exfiltration, and other threats targeting agentic AI systems.

Quick Start

Add this workflow to your repository at .github/workflows/scan-skills.yml:

name: Scan SKILL.md Files

on:
  pull_request:
    paths:
      - "**.md"

permissions:
  contents: read
  pull-requests: write

jobs:
  scan-skills:
    name: Malwar Skill Scan
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4

      - name: Scan SKILL.md files
        uses: Ap6pack/malwar/.github/actions/scan-skills@main
        with:
          path: "**/SKILL.md"
          fail-on: "SUSPICIOUS"

Inputs

Input	Description	Default	Required
path	Glob pattern for SKILL.md files to scan	**/SKILL.md	No
fail-on	Verdict threshold that causes the action to fail. One of: MALICIOUS, SUSPICIOUS, CAUTION	SUSPICIOUS	No
format	Output format: text, json, or sarif	text	No

Verdict Thresholds

Malwar assigns one of four verdicts to each scanned file:

Verdict	Risk Score	Description
CLEAN	0-14	No threats detected
CAUTION	15-39	Minor concerns found
SUSPICIOUS	40-74	Likely threats detected
MALICIOUS	75-100	Confirmed malicious content

The fail-on input controls when the action returns a non-zero exit code: - MALICIOUS — only fail on confirmed malware (most permissive) - SUSPICIOUS — fail on likely and confirmed threats (default) - CAUTION — fail on any concern (strictest)

Outputs

Output	Description
verdict	Worst verdict across all scanned files
risk_score	Highest risk score across all scanned files (0-100)
finding_count	Total number of findings across all scanned files
sarif_path	Path to the SARIF output file (only set when format is sarif)

Examples

Basic Usage

Scan all SKILL.md files and fail on suspicious or malicious content:

- name: Scan skills
  uses: Ap6pack/malwar/.github/actions/scan-skills@main

With SARIF Upload

Scan files and upload results to the GitHub Security tab for code scanning alerts:

- name: Scan skills
  id: malwar
  uses: Ap6pack/malwar/.github/actions/scan-skills@main
  with:
    format: sarif

- name: Upload SARIF
  if: always() && steps.malwar.outputs.sarif_path != ''
  uses: github/codeql-action/upload-sarif@v3
  with:
    sarif_file: ${{ steps.malwar.outputs.sarif_path }}
    category: malwar

Note: SARIF upload requires the security-events: write permission:

permissions:
  contents: read
  pull-requests: write
  security-events: write

Strict Mode

Fail on any finding, including low-risk cautions:

- name: Scan skills (strict)
  uses: Ap6pack/malwar/.github/actions/scan-skills@main
  with:
    fail-on: "CAUTION"

Custom Paths

Scan skill files in a specific directory:

- name: Scan vendor skills
  uses: Ap6pack/malwar/.github/actions/scan-skills@main
  with:
    path: "skills/vendor/**/SKILL.md"

Scan all Markdown files (not just SKILL.md):

- name: Scan all Markdown
  uses: Ap6pack/malwar/.github/actions/scan-skills@main
  with:
    path: "**/*.md"

Using Outputs in Subsequent Steps

- name: Scan skills
  id: malwar
  uses: Ap6pack/malwar/.github/actions/scan-skills@main
  with:
    fail-on: "MALICIOUS"

- name: Report results
  if: always()
  run: |
    echo "Verdict: ${{ steps.malwar.outputs.verdict }}"
    echo "Risk Score: ${{ steps.malwar.outputs.risk_score }}"
    echo "Findings: ${{ steps.malwar.outputs.finding_count }}"

Full Workflow Example

A complete workflow with SARIF upload and PR comments:

name: Scan SKILL.md Files

on:
  pull_request:
    paths:
      - "**.md"

permissions:
  contents: read
  pull-requests: write
  security-events: write

jobs:
  scan-skills:
    name: Malwar Skill Scan
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4

      - name: Scan SKILL.md files
        id: malwar
        uses: Ap6pack/malwar/.github/actions/scan-skills@main
        with:
          path: "**/SKILL.md"
          fail-on: "SUSPICIOUS"
          format: "sarif"

      - name: Upload SARIF to GitHub Security
        if: always() && steps.malwar.outputs.sarif_path != ''
        uses: github/codeql-action/upload-sarif@v3
        with:
          sarif_file: ${{ steps.malwar.outputs.sarif_path }}
          category: "malwar"

The action automatically posts a summary comment on pull requests with the scan verdict, risk score, and finding count. Existing comments are updated on subsequent runs rather than creating duplicates.