Developer-first automated LLM red-team platform
Project description
๐ฅ RedForge CLI
Developer-first automated LLM red-team platform
๐ Status: โ Sprint S-3 Complete (July 2025) - Product Hunt Ready, Enterprise Deployments Active
๐ Reference: Product One-Pager | 12M Roadmap | Landing Page
Problem We Solve
Large-language-model (LLM) apps ship to production with invisible jailbreak, data-leak, and cost-explosion risks. Regulators mandate continuous red-teaming (EU AI Act Art.55, US EO 14110).
RedForge automates OWASP LLM Top 10 testing:
- โ Local execution (keys on-prem)
- โ Audit-ready reports (NIST AI-RMF mapping)
- โ Docker CLI setup in 5 minutes
- โ Coverage tracking for compliance gaps
Example Use Cases
- Fintech Compliance: Scan GPT-4 for PCI DSS vulns, generate NIST-mapped PDF before audits.
- Enterprise Dev: Integrate OWASP tests in CI/CD, chaos-test for resilience.
- Security Research: Dry-run attacks, analyze feedback, customize for bias/PII threats.
Quick Start
pip install redforge
redforge doctor # Environment self-check
redforge scan gpt-4 --offline --dry-run
๐ Try Cloud Scan - $29/mo | ๐ Full Documentation | ๐ Threat Model
๐ณ Docker Alternative
docker run --rm siwenwang0803/redforge:latest scan gpt-4 --dry-run
โ๏ธ Cloud Starter ($29/month)
- Click Get Starter Key โ
- In CLI:
redforge signup --email you@example.com redforge scan gpt-4 --cloud-api-key <your_key>
- Access advanced features: unlimited scans, team collaboration, compliance reports
๐ฆ Advanced Installation
pip install redforge
redforge scan gpt-4 --dry-run
โ Helm (Kubernetes)
# Add repository
helm repo add redforge https://siwenwang0803.github.io/RedForge
helm repo update
# Install CLI for job-based scanning
helm install my-cli redforge/redforge-cli \
--set secrets.openaiApiKey="your-api-key"
# Install sidecar for runtime monitoring
helm install my-sidecar redforge/redforge-sidecar \
--set secrets.apiKeys.openai="your-api-key"
Core Features
โจ New in v0.3.1 โ Product Hunt Preview:
- ๐ฅ Open-core tiering: Free offline mode, Starter ($29/mo), Pro ($99/mo)
- ๐ Enhanced reporting: PDF/HTML/JSON with compliance mapping
- ๐ณ Stripe integration: Seamless checkout + webhook automation
- ๐ง Improved offline mode: No OPENAI_API_KEY required for dry runs
- ๐ Bug fixes: Typer 0.9 compatibility, CI/CD stability
- โ Production ready: Full E2E workflow validation
๐ Key Capabilities
- Automated Testing: OWASP LLM Top 10 (47 attacks), prompt injection/leakage detection, risk scoring (0-10 CVSS-like).
- Reporting: JSON/PDF/HTML/CSV with compliance (NIST, EU AI Act, SOC2, PCI DSS).
- Compliance & Audit: NIST AI-RMF/EU Act mappings, cryptographic trails.
- Production Ready: Docker/K8s, rate limiting, chaos testing, lightweight mode.
- Community-Driven: Feedback collection/analysis, automated roadmaps.
๐ Detailed Features (Click to Expand)
๐ Security Testing
- Complete OWASP LLM Top 10 coverage with 47 pre-built attacks
- Real-time vulnerability detection with confidence scoring
- Evidence collection and cryptographic audit trails
- Custom attack pack support (coming soon)
๐ Compliance & Reporting
- Multi-framework support: NIST AI-RMF, EU AI Act, SOC 2, ISO 27001, PCI DSS v4.0
- Executive summaries with risk assessments
- Remediation roadmaps and priority guidance
- Export to JSON, HTML, PDF, CSV formats
๐ Production Features
- Docker and Kubernetes deployment ready
- Rate limiting and timeout controls
- Chaos testing for resilience validation
- CI/CD integration with GitHub Actions
- Telemetry and feedback collection
Pro Features & Pilots
- Free OSS CLI for basics.
- Paid Pilots: $4-7k one-off pentests with custom PDFs (dev@redforge.com).
- Upcoming SaaS: $1k/mo for dashboards, monitoring, premium modules (FinOps, Privacy). Waitlist: redforge.com
Attack Packs
๐ด OWASP LLM Top 10 (Default)
| Category | Attacks | Severity | Description |
|---|---|---|---|
| LLM01 - Prompt Injection | 12 | Critical | Direct/indirect manipulation |
| LLM02 - Insecure Output | 6 | High | XSS/code injection |
| LLM03 - Training Data Poisoning | 4 | Medium | Data corruption attacks |
| LLM04 - Model DoS | 8 | High | Resource exhaustion |
| LLM05 - Supply Chain | 3 | Medium | Third-party vulnerabilities |
| LLM06 - Info Disclosure | 7 | Critical | Sensitive data leakage |
| LLM07 - Insecure Plugins | 5 | High | Plugin design flaws |
| LLM08 - Excessive Agency | 4 | Medium | Over-privileged actions |
| LLM09 - Overreliance | 3 | Low | Human dependency issues |
| LLM10 - Model Theft | 2 | Medium | IP extraction attempts |
Full details: Attack Packs Reference.
๐ฎ Coming Soon
- FinOps (cost detection, S-9)
- Privacy (GDPR/CCPA, S-10)
- Bias (fairness testing, S-11)
Configuration
Sample redforge.yaml:
target:
endpoint: "https://api.openai.com/v1/chat/completions"
model: "gpt-4"
scan:
max_requests: 100
timeout: 30
# Full config: See Configuration Reference
Roadmap & Status
- S-1 (โ Complete): CLI foundation, OWASP Top 10 coverage, report generation
- S-2/Pilot-0 (โ Complete): K8s sidecar, chaos testing, PCI DSS compliance
- S-3 (โ Complete - v0.3.1): Open-core model, Stripe payments, Product Hunt launch ready
- S-4 (๐ Next - Aug 2025): SaaS dashboard, team collaboration, advanced analytics
Full: 12M Roadmap
Development & CI/CD
Prerequisites: Python 3.11+, Poetry, Docker.
git clone https://github.com/siwenwang0803/RedForge.git
make install # Setup
make test # Run tests
CI/CD examples: CI/CD Guide.
Documentation & Links
- ๐ Full Documentation - Complete setup and usage guides
- ๐ก๏ธ Threat Model - Security architecture and risk analysis
- โ Helm Charts - Kubernetes deployment
- ๐ CLI Reference - Complete command documentation
- ๐ง Configuration Guide - Advanced configuration options
Support & Community
- Issues: GitHub Issues
- Cloud Support: dev@solvas.ai
- Enterprise: Schedule demo via landing page
Security & License
- Local exec, no exfil, opt-in telemetry.
- MIT License: LICENSE.
- Disclaimer: For authorized testing only.
Contributors
- AI Team: Claude 4 Sonnet (Dev Lead), ChatGPT o3-pro (Strategy), etc.
- Partners: 3 confidential enterprises.
Star History
Ready to secure your LLM? pip install redforge && redforge scan gpt-4 --dry-run
Questions: dev@redforge.com
SaaS Waitlist: redforge.com
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file redforge-0.3.1.tar.gz.
File metadata
- Download URL: redforge-0.3.1.tar.gz
- Upload date:
- Size: 86.1 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: poetry/2.1.3 CPython/3.13.5 Darwin/24.5.0
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
74bda233e89e87dc32860ad8b1618208fa31bb760d31b08e10da0cc13c0422ed
|
|
| MD5 |
8ef32013b97321c27f1a56be70dd776a
|
|
| BLAKE2b-256 |
586e24eacf6f1eff5708bb37c09fc0488cf1708207e5e109d51e59ed72f4eb09
|
File details
Details for the file redforge-0.3.1-py3-none-any.whl.
File metadata
- Download URL: redforge-0.3.1-py3-none-any.whl
- Upload date:
- Size: 95.0 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: poetry/2.1.3 CPython/3.13.5 Darwin/24.5.0
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
13c7d78a0643f6be7b18eb43eb21941403f272f6d1ddedcdea9437214d88a3f8
|
|
| MD5 |
dc77debb6e30ccb5798209cbce964b6a
|
|
| BLAKE2b-256 |
dac42969eceb918e1ecfb67e9440c64719f47391051e4783b50b54f71db0eaa4
|