whereismyai
Secret Guard
Secret Guard is a local LLM gateway that intercepts outbound requests, replaces sensitive data with realistic mocks, and restores real values in responses.
Secret Guard is a local LLM gateway designed to protect passwords, private keys, tokens, and cookies during AI agent workflows. It sits between developer tools and AI providers, intercepting outbound requests to replace secrets with equal-length realistic mocks before they leave the machine. When responses return, mocks are automatically restored to real values so tools function normally. It supports OpenAI, Anthropic, Gemini, Ollama, and Responses protocols with cross-protocol translation and multi-provider routing. Users can configure alternative strategies like rejection or canaries for sensitive credentials, and inspect all traffic through a built-in web user interface.
Secret Guard at a glance
Key features
- Replaces outbound secrets with realistic mock values
- Restores real values in incoming AI responses
- Supports OpenAI, Anthropic, Gemini, and Ollama protocols
- Provides reject and placeholder canary strategies
- Translates requests across different AI protocols
- Includes a web interface for session timeline inspection
- Executes plan-pool failover across multiple subscriptions
Use it for
- Protecting API keys and passwords from reaching AI vendors
- Running coding agents securely with real system credentials
- Routing requests across multiple AI providers and models
- Auditing AI traffic and inspecting session timelines
Alternatives to Secret Guard
- AgentGuard — Get a security and quality assessment of AI skills in seconds without an account.
- OpenAPPA — Tracks data flows algebraically to prevent prompt injection without breaking agents.
- Laya Guard — Judges every tool call locally on your laptop in ~0.1s with no per-decision billing.
- Portkey — Unified API access with observability, guardrails, and caching for LLM applications.
- Prediction Guard — Enforces NIST and OWASP policies on every model and agent from inside your security boundary.
- GitGuardian — Detects leaked secrets and insecure machine identities across your entire stack.
- AiSOC — Runs locally without API keys with bundled models and auditable reasoning ledgers.
- Repo Guards — It prevents AI agents from performing destructive actions across multiple repositories.
- Destructive Command Guard — It prevents AI agents from accidentally running commands that destroy work.
- Spring AI Privacy Guardrails — Protects sensitive data and secures tool usage within Spring AI applications.
- AgentGate — Blocks attacks in under 35ms with a local neural classification head.
- shim-cli — Masks sensitive data and tracks costs for coding agents locally.
All Secret Guard alternatives