Soroban Guard GitHub Action

Static analysis and security auditing for Soroban smart contracts. Integrates into GitHub Actions to detect reentrancy, arithmetic overflow, access control flaws, storage collisions, and other vulnerabilities before they reach production.

Table of Contents

• Features
• Prerequisites
• Quick start
• Usage
• Inputs
• Outputs
• Architecture
• Severity levels
• Integrations
• Contributing
• License

Features

• Automated scanning — runs on every push or pull request with no configuration beyond a workflow file.
• Inline annotations — findings are annotated directly on the affected lines in the diff view.
• PR summaries — a formatted comment with score, severity breakdown, and top findings is posted on each pull request.
• Security score (0–100) — quantifies overall contract health and tracks regressions over time.
• SARIF upload — feeds results into GitHub's Security > Code Scanning tab for historical tracking.
• Configurable thresholds — CI failure can be gated on severity level or score value.
• Monorepo support — multiple contract workspaces can be scanned in parallel using a build matrix.
• Exclusion rules — test fixtures, generated code, and third-party contracts can be skipped via glob patterns.
• Multi-format output — results can be produced in human-readable, JSON, or SARIF format.

Prerequisites

• A GitHub repository containing Soroban smart contracts written in Rust.
• GitHub Actions enabled on the repository.

Quick start

Create .github/workflows/audit.yml:

name: Security Audit
on: [pull_request]
jobs:
  audit:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: Soroban-Guard/Actions@v1
        with:
          path: contracts/

On every pull request, Soroban Guard scans contracts/ and fails the check if high-severity findings are present.

Usage

Basic

name: Security Audit
on: [pull_request]
jobs:
  audit:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: Soroban-Guard/Actions@v1
        with:
          path: contracts/

Full configuration

name: Security Audit
on: [pull_request, push]
jobs:
  audit:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: Soroban-Guard/Actions@v1
        id: guard
        with:
          path: |
            contracts/
            src/
          exclude: '**/test_*,**/fixtures/*'
          format: sarif
          min_severity: medium
          fail_on: high
          upload_sarif: true
          token: ${{ secrets.GITHUB_TOKEN }}

      - name: Check score
        run: |
          echo "Security score: ${{ steps.guard.outputs.score }}"
          if [[ "${{ steps.guard.outputs.score }}" -lt 70 ]]; then
            echo "Score below threshold!"
            exit 1
          fi

Monorepo setup

name: Security Audit
on: [pull_request]
jobs:
  audit:
    strategy:
      matrix:
        contract-path: [contracts/vault, contracts/nft, contracts/amm]
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: Soroban-Guard/Actions@v1
        with:
          path: ${{ matrix.contract-path }}

Pre-commit scan

name: Pre-commit Security Scan
on:
  pull_request:
    types: [opened, synchronize, reopened]
jobs:
  precommit:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
        with:
          fetch-depth: 0
      - uses: Soroban-Guard/Actions@v1
        id: guard
        with:
          path: contracts/
          format: json
          fail_on: critical

      - name: Post summary
        if: always()
        run: |
          echo "## Soroban Guard Summary" >> $GITHUB_STEP_SUMMARY
          echo "" >> $GITHUB_STEP_SUMMARY
          echo "- **Score**: ${{ steps.guard.outputs.score }}/100" >> $GITHUB_STEP_SUMMARY
          echo "- **Critical**: ${{ steps.guard.outputs.critical_count }}" >> $GITHUB_STEP_SUMMARY
          echo "- **High**: ${{ steps.guard.outputs.high_count }}" >> $GITHUB_STEP_SUMMARY

Fail conditions

fail_on value	CI fails when
critical	One or more critical findings exist
high	One or more high findings exist (default)
medium	One or more medium or higher findings exist
low	Any finding exists

Score-based gating can be added as a post-step:

- name: Enforce score threshold
  if: always()
  run: |
    SCORE=${{ steps.guard.outputs.score }}
    if [ "$SCORE" -lt 80 ]; then
      echo "Score $SCORE is below minimum 80"
      exit 1
    fi

Inputs

Input	Required	Default	Description
path	Yes	.	Path(s) to Soroban contract source files or directories. Supports newline-separated list for multiple paths.
format	No	sarif	Output format for report_path: human, json, or sarif. sarif is uploaded to code scanning when upload_sarif is enabled.
min_severity	No	high	Minimum severity to report: critical, high, medium, low, info. Findings below this threshold are omitted from annotations and comments.
exclude	No	''	Comma-separated glob patterns of files to skip (e.g. **/test_*,**/fixtures/*).
upload_sarif	No	true	Upload SARIF results to GitHub code scanning.
token	No	github.token	GitHub token for SARIF upload and PR comment posting.
fail_on	No	high	Severity level that causes the action to fail: critical, high, medium, low. Independent of min_severity.

Outputs

Output	Description
score	Overall security score (0–100). Higher values indicate fewer or less severe findings.
critical_count	Number of critical-severity findings.
high_count	Number of high-severity findings.
report_path	Filesystem path to the generated SARIF or JSON report.

Architecture

Soroban Guard is a Docker-based action built in two stages:

1. Builder stage — installs the published soroban-guard-core crate (v0.1.0) and copies the compiled Rust analyzer binary into a node:20-slim image alongside the Node.js orchestrator (src/main.js).
2. Runtime stage — entrypoint.sh invokes node src/main.js, which runs the analyzer in JSON mode to compute the score and findings, sets outputs, posts PR comments, creates inline annotations, generates a SARIF artifact for code scanning, and enforces the fail_on threshold.

The Node.js layer uses @actions/core, @actions/exec, and @actions/github for all GitHub API interactions.

Test fixtures

The fixtures/ directory contains sample Soroban contracts used by the test suite and CI: vulnerable contracts (token, amm_pair, escrow) that should produce findings, and a secure baseline (vault) that should pass. See fixtures/README.md.

Project ecosystem

Soroban Guard consists of three repositories that work together:

Repo	Description
Soroban-Guard/Actions	GitHub Action for CI/CD integration (this repo)
Soroban-Guard/Core	Rust static analyzer engine and CLI
Soroban-Guard/VS	VS Code extension for inline analysis

This action vs. the Core CLI: this repository is the turnkey CI integration — it bundles the analyzer binary from Core, so you never install Rust or wire up SARIF, comments, or annotations yourself. Use the Core CLI directly when you want to run scans from a terminal, a pre-commit hook, or a pipeline that doesn't use GitHub Actions.

Severity levels

Level	Description
critical	Exploitable vulnerabilities that can lead to loss of funds or contract compromise (e.g. reentrancy, missing access control).
high	Bugs likely exploitable or violating core security assumptions (e.g. arithmetic overflow, unsafe unwraps).
medium	Logic errors or design issues exploitable in edge cases.
low	Code quality or best-practice violations with limited security impact.
info	Suggestions and observations with no direct security impact.

Integrations

Code scanning

When upload_sarif: true (default), results are uploaded to GitHub's Security > Code Scanning tab, providing historical tracking and regression detection across branches.

PR comments

In pull request contexts, a summary comment containing the security score, severity breakdown, and top 10 findings is posted and updated on subsequent pushes.

Check annotations

Each finding is mapped to a GitHub check annotation on the affected file and line:

• critical / high → error
• medium → warning
• low / info → notice

Annotations appear directly in the Files changed tab of a pull request.

Contributing

See CONTRIBUTING.md for setup, style, and testing guidance. Source files for the action logic are in src/. The Rust analyzer source is maintained in the Soroban-Guard/Core repository.

License

MIT