Stop sending sensitive data to external AI providers.
API security proxy with reversible redaction. Scan every request for PII, secrets, and prompt injection — mask sensitive data on the way out, restore it on the way back. Open source. EU-hosted.
How it works
Detect
Grepture scans every request and response for PII, secrets, API keys, and sensitive patterns using configurable detection rules.
Act
Redact, mask, block, or log sensitive data based on your policies. Define rules per model, per endpoint, or per data type.
Rehydrate
On the response path, Grepture restores redacted tokens so your application receives complete, usable data.
Works with
Zero retention
Full protection. Zero stored data.
Enable zero-data mode and Grepture processes every request — detecting PII, redacting secrets, blocking threats — without ever writing your content to disk. Headers, bodies, and URLs never touch our database. Only operational metadata is logged.
- Rules still fire — PII detection, redaction, blocking, and tokenization all work in-flight
- Only method, status code, latency, and rule hits are stored
- One toggle in your dashboard. Instant. No migration needed.
Capabilities
Find personal data automatically
Names, emails, phone numbers, SSNs, addresses — detected and handled before they leave your infrastructure.
Catch leaked credentials
API keys, tokens, passwords, and connection strings are identified and redacted before they reach any external service.
Guard proprietary content
Prevent source code, internal documents, and trade secrets from being sent to third-party services.
Block adversarial inputs
Detect and neutralize prompt injection attempts before they reach your LLM, protecting your AI pipeline.
Built for trust
Your data never leaves Europe.
All Grepture infrastructure runs in the EU. Every subprocessor — database, cache, analytics, payments — is hosted in Germany or Ireland. GDPR-ready by default.
- +All infrastructure hosted in Frankfurt & Nuremberg
- +Every subprocessor EU-based — no US data transfers
- +GDPR and EU AI Act ready out of the box
- +Zero-data mode: nothing written to disk
Don’t trust a black box with your data.
The Grepture proxy is fully open source. Every detection rule, every redaction action, every byte of data handling is auditable. Self-host for full infrastructure control.
- +Full proxy source code on GitHub
- +Every detection rule readable and auditable
- +No black-box processing — deterministic and transparent
- +Self-host option for teams that need full control
Your command center
Real-time visibility into every request.
Inspect every request, see which rules fired, and drill into headers and bodies.
Use cases
AI-Powered Apps
Chatbots, summarizers, and any app making LLM calls. Mask PII and secrets on the way out, restore them on the way back — your users never notice, the model never sees real data.
AI Agents
Agents with tool access, multi-step workflows, and MCP servers. The proxy sits on the network path — no matter what the agent does, sensitive data stays protected.
RAG Pipelines
Knowledge bases pulling internal docs often contain secrets, PII, and proprietary content. Scan every chunk before it reaches the model.
Multi-Model Security
Teams using multiple AI providers need one security layer. One proxy, consistent policies, unified audit trail — across every model and provider.
Any API Call
Not just AI — wrap any outbound HTTP call with grepture.fetch() and scan for sensitive data before it leaves your network.
Start protecting your API traffic in 5 minutes
Deploy Grepture in minutes. No code changes required.
Free for up to 1,000 requests/month · No credit card required
Get Started Free