promptstrike 0.2.0a0

Developer-first automated LLM red-team platform

🎯 PromptStrike CLI

Developer-first automated LLM red-team platform

🚀 Status: ✅ Sprint S-3 Preparation (July 2025) - S-2/Pilot-0 Complete, Enterprise Ready
📋 Reference: Product One-Pager | 12M Roadmap | Changelog | DOD Summary

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).

PromptStrike automates OWASP LLM Top 10 testing:

• ✅ Local execution (keys on-prem)
• ✅ Audit-ready reports (NIST AI-RMF mapping)
• ✅ Docker CLI setup in 5 minutes
• ✅ 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

🐳 Docker (Recommended)

# Clone and build locally
git clone https://github.com/siwenwang0803/PromptStrike.git
cd PromptStrike
docker build -t promptstrike/cli .

# Run scan
docker run --rm -e OPENAI_API_KEY=$OPENAI_API_KEY -v $(pwd)/reports:/app/reports promptstrike/cli scan gpt-4

Docker Hub image coming soon - currently build locally.

📦 PyPI

pip install promptstrike
promptstrike scan gpt-4 --dry-run

Core Features

New in v0.2.0: Guardrail side-car for K8s, enhanced chaos testing (resilience scoring), community feedback system.

• 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.

📄 Detailed Features (Click to Expand)

🔒 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)

📊 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

🚀 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 ([e-mail]).
• Upcoming SaaS: $1k/mo for dashboards, monitoring, premium modules (FinOps, Privacy). Waitlist: promptstrike.com

Attack Packs

🔴 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.

🔮 Coming Soon

• FinOps (cost detection, S-9)
• Privacy (GDPR/CCPA, S-10)
• Bias (fairness testing, S-11)

Configuration

Sample promptstrike.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 (Shipped): CLI, OWASP coverage, reports.
• S-2/Pilot-0 (Complete): K8s sidecar, chaos enhancements, PCI DSS. Target: Enterprise deployment.
• Upcoming: S-3 (pilots/Stripe, $15k revenue), S-4 (SaaS dashboard).

Full: 12M Roadmap.

Development & CI/CD

Prerequisites: Python 3.11+, Poetry, Docker.

git clone https://github.com/siwenwang0803/PromptStrike.git
make install  # Setup
make test     # Run tests

CI/CD examples: CI/CD Guide.

Support & Community

• Issues: GitHub
• Docs: CLI Spec, API (Soon)
• Community: Discord/Slack (Soon)

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 promptstrike && promptstrike scan gpt-4 --dry-run

Questions: [e-mail]
SaaS Waitlist: promptstrike.com