BotGuard Security Scan - GitHub Action

Automated security scanning for AI agents in your CI/CD pipeline. Tests for prompt injection, jailbreaks, data extraction, and 70+ attack scenarios aligned to OWASP LLM Top 10.

Quick Start

name: AI Security Scan
on: [pull_request]

jobs:
  security:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4

      - uses: botguardai/security-scan@v1
        with:
          api-key: ${{ secrets.BOTGUARD_API_KEY }}
          endpoint: https://my-agent.com/chat
          description: "Customer support chatbot"
          fail-threshold: 80

Inputs

Input	Required	Default	Description
api-key	Yes	—	BotGuard API key
endpoint	No	—	Agent API endpoint URL
description	No	AI agent	Agent description
system-prompt	No	—	System prompt (text or file path)
fail-threshold	No	0	Minimum score to pass (0 = never fail)
categories	No	—	Comma-separated attack categories
attack-count	No	—	Number of attacks to run
format	No	table	Output format: table, json, summary
comment	No	true	Post results as PR comment
api-url	No	—	Custom API URL (self-hosted)

Outputs

Output	Description
score	Security score (0-100)
passed	Whether threshold was met (true/false)
report-url	Link to full security report
session-id	Scan session ID

Examples

Block PRs Below Security Threshold

- uses: botguardai/security-scan@v1
  with:
    api-key: ${{ secrets.BOTGUARD_API_KEY }}
    system-prompt: ./prompts/system.md
    fail-threshold: 80

Scan Specific Attack Categories

- uses: botguardai/security-scan@v1
  with:
    api-key: ${{ secrets.BOTGUARD_API_KEY }}
    endpoint: https://my-bot.com/chat
    categories: "jailbreak,prompt_injection,data_extraction"
    attack-count: 50

Use Score in Downstream Steps

- uses: botguardai/security-scan@v1
  id: scan
  with:
    api-key: ${{ secrets.BOTGUARD_API_KEY }}
    endpoint: https://my-bot.com/chat

- run: echo "Score is ${{ steps.scan.outputs.score }}"

Scan on Schedule

name: Weekly Security Scan
on:
  schedule:
    - cron: '0 9 * * 1'

jobs:
  scan:
    runs-on: ubuntu-latest
    steps:
      - uses: botguardai/security-scan@v1
        with:
          api-key: ${{ secrets.BOTGUARD_API_KEY }}
          endpoint: https://my-bot.com/chat
          fail-threshold: 70

PR Comment

When comment: true (default), the action posts a summary on pull requests:

Metric	Value
Score	85/100
Passed	42
Failed	8
Violations	12

Getting Started

1. Sign up at botguard.dev
2. Go to Account > API Keys and create a CI/CD key
3. Add the key as a GitHub secret: BOTGUARD_API_KEY
4. Add the action to your workflow

License

MIT