// THE PLATFORM
CENTRAL HUB
🦞 meda-claw
The Independent AI Governance & Security Stack — the open-source alternative to proprietary AI evaluation platforms. Unified CLI orchestrating 8 tools: behavioral auditing, IP provenance with Human-Review Attestation (Policy-as-Code), API permission scanning, AST analysis, LLM evaluation, browser endpoint security, notification sanitization, and LocalFirstAI integration. One command: pip install meda-claw.
PythonCLIAI GovernancePolicy-as-CodeSecurityPlatform
COMPLETE
🍋 LocalFirstAI
Privacy-first local AI agent framework — zero cloud, full control. AMD Lemonade + Google Gemma 4 for 100% local inference. Agent framework with 6+ tools, SQLite memory, privacy monitor/auditor/enforcer, YAML plugin system, and FastAPI web dashboard. ~8K lines Python.
PythonAMD LemonadeGemma 4FastAPISQLitePrivacyLocal AI
// THE AI GOVERNANCE SUITE
FLAGSHIP — AI GOVERNANCE
🔐 git-provenance
AI Governance & IP Compliance Firewall for Git. Enforces provenance tracking, AI contribution gating (configurable thresholds), model allow-listing, and generates structured audit reports for legal/compliance teams. Pre-commit hook integration for CI/CD pipelines.
PythonGit HooksAI GovernanceIP ComplianceAudit
FLAGSHIP — AI GOVERNANCE
🔑 API Auditor
Real-time CLI + CI/CD auditor for API key permissions. Multi-provider scanning (Google, AWS, Stripe), live validation against provider APIs, Gemini/AI key flagging, policy drift detection, and a React-powered analytics dashboard.
PythonReactCLICI/CDSecurity
FLAGSHIP — AI GOVERNANCE
⚡ ProjectSpark
AI Governance & Evaluation Platform — LLM benchmark sensitivity analysis across 5 prompt architectures (zero-shot, chain-of-thought, persona, few-shot, delimiter). CLEAR Act copyright compliance dashboard with federal disclosure PDF generation.
PythonFastAPIReactDockerAI GovernanceLLM Eval
FLAGSHIP — AI GOVERNANCE
🔍 Agent Audit
Behavioral Audit Engine for autonomous AI agents. Real-time forensic monitoring of Codex, Claude Code, and local OS-level agents. Multi-factor risk scoring (RAG), tamper-evident SHA-256 hash chain, configurable security policies, secret detection, and C-suite executive risk dashboards.
PythonAgent SecurityForensicsRisk ScoringAudit
// THE FORENSICS SUITE
FLAGSHIP — FORENSICS
🔬 Repo X-Ray
AST-powered security scanner + interactive D3.js dependency graph visualizer. Detects exposed secrets, maps codebase architecture, and ranks files by PageRank for attack surface analysis. OSINT-grade data disclosure detection.
AST AnalysisD3.jsOSINTSecurityGraph Visualization
// PRIVACY & DETECTION
IN DEVELOPMENT
📱 DeepfakeShield
On-device deepfake detection for mobile. Share a suspicious image from any app → get a forensic analysis with 6 checks (face boundaries, lighting, compression artifacts). React Native (Expo) with Web support, fully offline, privacy-first. ML model training in progress.
React NativeExpoExpo WebTFLiteComputer VisionPrivacy
COMPLETE (v1.0)
👁️ ShadowProfile Elite
Privacy-first adversarial audit platform. 3-agent pipeline (Harvester → Profiler → Adversary) with local LLM (Ollama/Llama 3.2). 3D data cloud (Three.js), exposure radar, spear-phishing simulation. 100% local, zero telemetry.
PythonOllamaThree.jsPlaywrightChromaDBPrivacy
COMPLETE (v1.0)
📜 ConsentLedger Sovereign
AI opt-out enforcement platform. C2PA content signing, zero-width canary tokens, .well-known/ai-consent.json (EU AI Act Art. 4(3)). Forensic evidence PDF + cease & desist generator. Self-hosted with Docker Compose.
PythonC2PAPyTorchDockerEU AI ActPrivacy
// SECURITY TOOLS
PUBLISHED — CHROME WEB STORE
🛡️ ExtensionGuard
Browser Safety Suite — 5 security modules protecting you in real-time. Extension monitor (A-F grading), clipboard sanitizer, phishing URL shield, clickjacking detector, AI privacy filter. 100% local processing.
🔍 Extension Monitor
📋 Clipboard Sanitizer
🔗 URL Shield
👻 Ghost Script Monitor
🤖 AI Privacy Filter
JavaScriptChrome MV3SecurityPrivacy
ACTIVE — SECURITY
🔔 Push Guardian
Enterprise-grade middleware for push notification sanitization. Safety-Net protocol with Fail-Open guarantee. Chrome Extension (MV3) that intercepts, analyzes, and filters malicious push payloads before they reach the user.
Chrome MV3MiddlewareSecurityEnterprise
// PROMPT ENGINEERING
ACTIVE — CLI TOOL
⚒️ PromptForge
Local-first, model-aware prompt optimizer. Applies 5 optimization passes (compress, structure, chain-of-thought, guardrails, model-adapt) to measurably improve prompt quality and token efficiency. Model-aware: Claude→XML tags, GPT→Markdown, Llama→concise. Fills the vacuum left by PromptPerfect's shutdown (Sept 2026). No SaaS. No cloud dependency.
Python
CLI
Prompt Engineering
Claude
GPT-4
Local-First
// CONSUMER TOOLS
PUBLISHED — CHROME WEB STORE
🔮 Size-Oracle
Universal size advisor — Input measurements once, get confidence-scored recommendations on any clothing site. Weighted matching algorithm across 100+ retailers. Privacy-first, all processing local.
JavaScriptChrome ExtensionData AnalyticsPrivacy
// CORE UTILITIES
ACTIVE — AI TOOLS
🧠 AI Brain Vault
Local-first AI conversation capture & semantic search. Chrome extension auto-captures ChatGPT, Claude, and Gemini conversations. FastAPI backend with sentence-transformers for semantic similarity search. Zero cloud, maximum privacy.
PythonFastAPIChrome ExtensionNLPSQLiteSentence-Transformers
CLASSIFIED
🔒
CLASSIFIED
// NEXT WEAPON LOADING...