What is SecureAIFlow?

SecureAIFlow is a free VS Code extension and web chat that redacts API keys, secrets, and PII from your prompts before they reach AI providers like ChatGPT, Claude, or Gemini. Install takes 30 seconds and works alongside GitHub Copilot, Cursor, and other AI tools.

Is SecureAIFlow really free?

Yes. The Free tier includes the VS Code extension, web chat with 10 messages/day on Gemini Flash, and supports up to 3 users per team. No credit card required. Business email required for signup.

How does SecureAIFlow detect secrets and PII?

SecureAIFlow uses a three-stage hybrid detection pipeline (regex + fine-tuned MiniLM + HMAC-SHA256 pseudonymization) achieving an F1 score of 0.976. Detected items are replaced with deterministic pseudonyms that preserve code structure.

Where is my data processed?

SecureAIFlow is deployed on-premises, on your own infrastructure. Prompts are processed in-memory only — nothing is stored. Data never leaves your environment. DPA and subprocessor list available publicly.

Does it work with GitHub Copilot and Cursor?

Yes. SecureAIFlow is a complementary security layer. Use Copilot or Cursor for autocomplete, and use SecureAIFlow's chat for refactors, explanations, or any task involving sensitive code. They run side by side.

How much does SecureAIFlow cost?

Free tier: €0 forever, up to 3 users. BYOK Security Layer: €16 per seat per month (minimum 5 users). Managed Smart Routing: €30 per seat per month with AI usage included. Enterprise on-premise: custom pricing starting from €60 per seat.

Use ChatGPT, Claude, Copilot at work. Without the guilt.

SecureAIFlow strips API keys, secrets, and PII from your prompts before they leave your IDE. Self-hosted on your infrastructure. GDPR & DORA ready.

Install for VS Code — Free Claude Code — Zero config Chrome/Edge explorer Plugin

Free forever No credit card 30-second install

For teams of 5+ or enterprise? Book a demo →

How your prompts flow

Your IDE / Browser

apiKey = "sk-abc..."

→

SecureAIFlow

On-prem · in-memory · no storage

→

Claude · GPT · Gemini

apiKey = "[REDACTED_1]"

Your prompts never touch disk. Redaction happens in memory, on your infrastructure, in milliseconds. Nothing is stored, nothing is trained on.

Live Now · VS Code Marketplace

Secrets redacted before they leave your IDE

API keys, credentials, and tokens are detected and replaced with pseudonyms directly in your VS Code workflow. Chat, refactor, explain — sensitive data never reaches OpenAI, Anthropic, or Google.

Install from VS Code Marketplace →

New · Terminal-first

Claude Code, proxied. Zero config.

Two environment variables — no install, no code change. SecureAIFlow proxies every Claude Code request through the redaction engine. Bash outputs, file reads, and tool_results are sanitized before reaching Anthropic. Bring your own Anthropic key, keep your audit trail.

See the setup →

# 1. Point Claude Code at SecureAIFlow

export ANTHROPIC_BASE_URL=https://api.secureaiflow.com

export ANTHROPIC_CUSTOM_HEADERS="x-saf-user: [email protected]"

# 2. Use Claude Code as usual

$ claude "fix the failing auth tests"

↳ tool_result · cat .env

DB_PWD=prod-s3cret!2026

DB_PWD=[REDACTED_1] ✓ sent to Anthropic

Live · Chrome Web Store

Keep using ChatGPT and Claude — just safely

For teams attached to chat.openai.com or claude.ai directly. Our Chrome plugin intercepts and redacts prompts inside the ChatGPT / Claude / Gemini web UIs before they're sent. One click to install from the Chrome Web Store.

Install from Chrome Web Store → Questions?

For Leadership

Full visibility — the control plane your CISO wants

Secrets caught, tokens scanned, cost per LLM provider, team activity. Every AI interaction tracked and measured — with a full audit trail that satisfies DORA Article 28 and GDPR requirements.

The Problem

Every AI prompt is a potential data leak

API keys, credentials, PII — pasted into ChatGPT, Copilot, and every AI tool. You can't block AI. Your devs will use it anyway. You need to secure it.

Without SecureAIFlow

✕API keys sent to ChatGPT in plain text
✕Customer PII shared with AI models
✕No visibility on what's leaving your org
✕DORA Article 28 & GDPR exposure
✕Devs use personal accounts to bypass IT

With SecureAIFlow

+Secrets detected and redacted in real-time
+PII pseudonymized before reaching any LLM
+Full audit trail of every AI interaction
+GDPR & DORA compliant from day one
+Devs keep using their favorite AI tools

How It Works

Three steps. Secrets stay yours.

SAF sits between your teams and every AI tool they use.

Intercept

The VS Code extension , Chrome plugin and web chat capture prompts before they leave your environment.

Detect & Redact

Three-stage hybrid pipeline (regex + fine-tuned MiniLM + deterministic pseudonymization) reaches F1 0.97. Connect your vault for 100% coverage.

Route to LLM

BYOK mode sends to your provider. Smart Routing intelligently routes across Claude, GPT-5, and Gemini based on complexity and cost.

Enterprise Deployment

On-Premises Only

Self-hosted security layer deployed on your infrastructure. Complete control, full compliance, custom pricing.

Enterprise

On-Premises Deployment

Self-hosted on your infrastructure. Complete sovereignty and control.

Custom Pricing

Tailored to your organization

Deployed on your infrastructure (K8s, VMs, bare metal)
Real-time secret detection (97% F1-score)
VS Code extension, CLI, Dashboard, API
SSO (SAML, Okta) & multi-user management
Complete audit trail & compliance reporting
SLA 99.95% with dedicated support
Custom integrations & guided POC

Schedule Demo Contact Sales

Use ChatGPT, Claude, Copilot at work. Without the guilt.