Skip to content

AI Security Summary

Generates an AI-powered executive security summary from aggregated Argus scanner results

uses: huntridge-labs/argus/.github/actions/ai-summary@0.6.7

Generates an AI-powered executive security summary from aggregated Argus scanner results. Consumes scanner-summary-* artifacts produced by other Argus scanner actions and delivers a structured summary as a GitHub Issue and/or PR comment.

Supported AI Providers

Provider Input Required
GitHub Copilot COPILOT_GITHUB_TOKEN
Anthropic Claude ANTHROPIC_API_KEY
Google Gemini GEMINI_API_KEY

Usage

- uses: huntridge-labs/argus/.github/actions/ai-summary@0.6.7
  env:
    ANTHROPIC_API_KEY: ${{ secrets.ANTHROPIC_API_KEY }}
    GITHUB_TOKEN:      ${{ secrets.GITHUB_TOKEN }}
  with:
    provider: 'claude'
    post_issue: 'true'

Inputs

Input Description Required Default
provider AI provider to use: copilot, claude, or gemini No copilot
max_findings Maximum findings to include per scanner in the prompt No 20
post_issue Post the summary as a GitHub Issue No true
post_pr_comment Post the summary as a PR comment No false
issue_label Label to apply to the generated GitHub Issue No security-summary
fail_on_ai_error Fail the workflow if AI summary generation fails No false
skip_download Skip internal artifact download when summaries are pre-populated by the caller No false

Outputs

None. Results are delivered as a GitHub Issue and/or PR comment.

Required Permissions

permissions:
  contents: read
  issues: write
  pull-requests: write  # only needed if post_pr_comment: true

Secrets

Secret Required For
COPILOT_GITHUB_TOKEN GitHub Copilot provider
ANTHROPIC_API_KEY Anthropic Claude provider
GEMINI_API_KEY Google Gemini provider
GITHUB_TOKEN Posting Issues and PR comments

Example: Full Pipeline Integration

Add ai-summary as the final step after your scanners have run:

jobs:
  # ... your scanner jobs ...

  scanner-bandit:
    uses: huntridge-labs/argus/.github/actions/scanner-bandit@0.6.7
    ...

  scanner-gitleaks:
    uses: huntridge-labs/argus/.github/actions/scanner-gitleaks@0.6.7
    ...

  ai-summary:
    needs: [scanner-bandit, scanner-gitleaks]
    runs-on: ubuntu-latest
    permissions:
      contents: read
      issues: write
    steps:
      - uses: actions/checkout@v6
      - uses: huntridge-labs/argus/.github/actions/ai-summary@0.6.7
        env:
          ANTHROPIC_API_KEY: ${{ secrets.ANTHROPIC_API_KEY }}
          GITHUB_TOKEN:      ${{ secrets.GITHUB_TOKEN }}
        with:
          provider: 'claude'
          post_issue: 'true'

Manual Execution

The AI Scan Summary workflow in the Actions tab allows you to generate a summary on demand for any PR that has a completed security scan run. Select the AI provider and enter the PR number — the workflow resolves the scan artifacts automatically.

Summary Structure

The generated summary includes:

  • Key Findings Summary — table of finding counts by severity
  • Executive Overview — 1-2 sentence posture summary
  • Critical Risk Areas — detailed breakdown of critical/high findings
  • Risk Assessment — overall risk level with rationale
  • Recommended Actions — prioritized remediation steps with timelines
  • Compliance Considerations — relevant frameworks (NIST, FedRAMP, etc.)
  • Appendices — scanning tools used, affected containers