Skip to content

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

270 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

SecGuardian

AI-Native Security Workflow for the Entire Software Development Lifecycle

Platform Version Go Detectors CWE Top 25

SecGuardian is an enterprise-grade AI application security framework. It integrates AI reasoning with security engineering practices to help teams build, review, fix, and release secure software through four automated security gates.

By moving beyond raw AI code generation and introducing Rule Packs, SecGuardian translates standards like OWASP ASVS, NIST SSDF, and PCI DSS into machine-executable audit rules that run in your CLI and CI/CD pipelines.


Quick Start

Choose how you want to install:

Plugin zip (recommended)

Download the plugin zip for your AI agent from Releases:

If you use... Download this file
Claude Code secguardian-0.10.0-claude-code-<os>-<arch>.zip
OpenCode secguardian-0.10.0-opencode-<os>-<arch>.zip
Gemini CLI secguardian-0.10.0-gemini-cli-<os>-<arch>.zip

Extract the zip into your agent's plugin directory, or install it through your agent's extension manager. Then use:

/secguard ./src           # Secure coding guidance
/secreview                # Code review (auto git diff)
/secfix                   # Auto-remediation
/secaudit                 # Full audit

Source tarball (with deploy.sh)

Download secguardian-0.10.0-source.tar.gz from Releases, extract, then:

bash scripts/deploy.sh          # Install for all supported platforms
bash scripts/deploy.sh cc       # Claude Code only
bash scripts/deploy.sh nga      # OpenCode only
bash scripts/deploy.sh cac      # Gemini CLI only

Build from source (development)

git clone https://github.com/DannyAn/SecGuardian.git
cd SecGuardian
bash scripts/dev-deploy.sh

Try the demo

After deploying, run against the built-in vulnerable examples:

/secguard examples/cpp-vuln-demo/src cpp
/secreview examples/python-vuln-demo/src python
/secaudit --rulepack secguardian examples/java-vuln-demo/src java

Standalone indexer (Go developers)

cd internal && go build -o secguardian-index .
./secguardian-index --path ./src --output index.json

The Secure SDLC Pipeline

The Software Development Lifecycle (SDLC) describes the stages of building software — from writing code and reviewing changes to shipping a release. SecGuardian embeds security into each of these stages. Rather than a single monolithic scan, it provides contextual tools purpose-built for each phase.

graph TD
    A[Writing Code] -->|/secguard - Prevent| B(Committed Code)
    B --> C{Pull Request}
    C -->|/secreview - Detect| D[Security Findings]
    D -.->|/secfix - Remediate| C
    D -->|Approved| E(Merge)
    E --> F{Release Candidate}
    F -->|/secaudit - Verify| G[Evidence & Reports]
    G -->|Pass| H((Release))

    classDef gate fill:#2b3137,stroke:#00ADD8,stroke-width:2px,color:#fff;
    class D,G gate;
Loading

The Four Security Gates

Command Stage Primary User Outcome
/secguard Implementation Developers Continuous secure coding guidance during development.
/secreview Pull Request Reviewers Context-aware AI code review using business logic and exploitability analysis.
/secfix Remediation Developers Auto-generates deterministic, ready-to-apply patches from findings.
/secaudit Release Security / Ops Evaluates the application against enterprise Rule Packs before release.

These four commands represent different stages of Secure SDLC, not different levels of scanning.


Enterprise Governance as Code (Rule Packs)

Raw AI reasoning is powerful, but enterprise security requires consistency, traceability, and compliance. SecGuardian achieves this through Rule Packs — machine-executable knowledge assets that map published standards to audit rules, evidence requirements, and pass/fail criteria.

audit-framework/
├── rulepacks/             # Pluggable rule pack definitions
│   ├── secguardian/       # Built-in default (17 audit domains)
│   │   ├── pack.json      # Manifest + standard mappings
│   │   └── rules/         # 17 audit rules
│   ├── company-redline/   # Future: enterprise security baseline
│   └── owasp-asvs/        # Future: OWASP ASVS compliance
├── engine/                # Execution engine specification
├── templates/             # Report template definitions
└── reporters/             # Output format extensions

Teams can version, review, and customize Rule Packs independently of the AI model. The AI engine can be upgraded without touching the security knowledge, and vice versa — Rule Packs outlast the AI model.

Current coverage

CWE Top 25: 100% coverage. OWASP Top 10: 100% coverage.

60 detectors across 7 security namespaces, supporting C/C++/Java/Python/Go/JavaScript:

Namespace Count Coverage
memory 13 Buffer overflow, UAF, double-free, null dereference
concurrency 4 Race conditions, deadlocks
system 7 Command injection, path traversal
crypto 9 Weak algorithms, hardcoded keys
web 21 XSS, SQLi, SSRF, CSRF, prototype pollution
resource 6 File/socket/resource leaks, lock misuse
error 6 Stack trace leakage, sensitive data in logs

Analysis strategies include Taint Analysis, Data Flow Analysis, Attack Surface Analysis, Trust Boundary Analysis, and State Machine Analysis. These are tools, not the product — the long-term value lies in the Rule Packs that define what to analyze and the workflow that acts on the results.

Standards in development

  • OWASP ASVS (Application Security Verification Standard)
  • NIST SSDF (Secure Software Development Framework)
  • CIS Benchmarks
  • PCI DSS
  • Enterprise Security Baselines

Developer Experience

SecGuardian is designed for zero friction in daily use. All commands work with zero arguments — they auto-detect the target language, current directory, and most recent scan context.

Automated Remediation

Reduce a 30-minute vulnerability triage cycle to roughly two minutes:

# 1. Detect vulnerabilities in your current PR
/secreview

# 2. Generate ready-to-apply patches
/secfix findings/web-sql-injection/

# 3. Review the patch and apply
git diff
git apply findings/web-sql-injection/*.patch

# 4. Re-run to verify
/secreview

SecGuardian never merges code automatically. The developer retains final decision-making authority on every patch.

CI/CD Integration

# .github/workflows/security.yml
- uses: secguardian/secguardian-action@v1
  with:
    mode: secaudit
    skill: taint-analysis
    path: src/

Also supports GitLab SAST (artifacts:reports:sast) and Azure DevOps SARIF upload.

Scan Output

.codeagent/<extension>/scans/<scan-id>/
├── human/executive-summary.md     # Executive dashboard
├── findings/                      # Per-detector finding files
├── ai/remediation-pack.json       # AI-consumable remediation pack
├── report.md                      # Human-readable report
├── dashboard.html                 # Management dashboard
├── results.sarif                  # SARIF 2.1.0 (CI/CD input)
├── summary.json                   # Summary statistics
├── manifest.json                  # Scan metadata + finding index
├── status.json                    # CI gate status
└── delta.json                     # Delta vs previous scan

Vision

Our vision is to make enterprise-grade application security accessible to every development team. Security engineers remain at the center of the process — they work with better tooling, not fewer people.

We are building toward that vision in three phases:

AI Security ScannerAI Security WorkflowAI Security Governance Platform

The Audit Framework and Rule Packs we are building today lay the foundation for that third phase: a governance platform where organizations define their security policies as executable rules, run them across every stage of development, and produce audit-ready evidence for every release.


License

Proprietary. All rights reserved.

About

No description, website, or topics provided.

Resources

Security policy

Stars

4 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages