Frontier-grade Claude models with agentic workflows, strong coding, and enterprise guardrails delivered via developer console, web app, and cloud partners.
Corrath
FreeStarting price
0Popularity
About Corrath
Corrath is a deterministic security gateway positioned between applications and AI providers such as OpenAI, Claude, and Gemini. It neutralizes prompt injection attacks by filtering malicious inputs before they reach the provider. The platform enforces per-developer spending limits to prevent cost overruns and reroutes requests to alternative providers during outages in under 5 milliseconds. Corrath includes a semantic firewall, cost forecasting tools, HMAC-signed webhooks for secure callbacks, real-time usage auditing, and multi-provider routing, all accessible through a unified dashboard. It is designed for teams that need reliable, secure, and cost-controlled access to AI models without managing infrastructure themselves. The system is particularly useful for organizations that rely on multiple AI services and require consistent performance and security across providers.
Key features
- Semantic firewall to block prompt injection attacks
- Per-developer spending limits and cost forecasting
- Multi-provider routing with automatic failover
- Real-time usage auditing and reporting
- HMAC-signed webhooks for secure callbacks
- Unified dashboard for monitoring and management
- Sub-5ms failover during provider outages
- Freemium pricing with a free tier of 500 requests/month
Use cases
- Securing AI-powered applications against prompt injection vulnerabilities
- Controlling and forecasting AI usage costs across teams or projects
- Ensuring continuous AI service availability during provider outages
Pros
- Neutralizes prompt injection attacks and adversarial manipulations before they reach AI providers
- Enforces granular spending limits per developer key to prevent cost overruns
- Provides sub-5ms failover routing during provider outages with automated circuit breakers
- Includes a semantic firewall and in-memory vector scanner for advanced threat detection
- Offers real-time usage auditing, cost forecasting, and HMAC-signed webhooks for secure callbacks
Cons
- Requires integration with Corrath's gateway layer, adding a proxy step between applications and AI providers
- Enterprise-grade features may introduce additional complexity for smaller teams or simpler use cases
Frequently asked questions about Corrath
How does the security firewall scan prompts?
The semantic firewall uses an in-memory high-performance vector scanner with embeddings to detect adversarial injections, jailbreak attempts, and PII leaks before forwarding requests to AI providers.
Do you store or log completion prompts?
Corrath anonymizes and logs metrics such as spend, latency, and threat categories, but does not store or log the actual completion prompts to protect user privacy.
How fast is the automatic failover router?
The automatic failover router reroutes completion requests in under 5 milliseconds during provider outages, ensuring minimal disruption to applications.
Can I enforce developer spending limits?
Yes, Corrath allows enforcing granular spending limits per developer key, automatically blocking excess completion calls to prevent budget overruns.
What AI providers does Corrath support?
Corrath supports dynamic routing between major providers including OpenAI, Claude, Gemini, Mistral, and DeepSeek with standardized syntax.
Is there a free tier available?
Yes, Corrath offers a free tier with essential completions limits, basic proxy gateway, security firewall scanning, and provider outage fallbacks for testing and small-scale use.
