$ loading_
Let AI agents use secrets safely for server-side HTTP, shell, and SSH actions.
Copy the install command and let the AI configure it · recommended for beginners
No copy-paste install info for "safe-key-mcp" yet — see the docs or source repo.
Use the API key stored in the encrypted vault to send an HTTP request server-side, and return only the API response without exposing the secret value.
The server completes the API call and returns the result without exposing the secret in plaintext to the AI.
Use SSH credentials stored in the vault to connect to the target server, execute the specified command server-side, and return the command output and status.
It returns the remote command result while keeping the credentials server-side and hidden from the AI.
Run a shell script that requires an access token, reading the token from the encrypted vault and executing the script server-side with a summarized result.
The script runs on the server and returns a task summary or log excerpts without revealing the token.
When developers or DevOps engineers need an AI agent to call authenticated external APIs, they can store secrets in an encrypted vault and let the server make the request without exposing the secret to the model.
For operational tasks that require shell or SSH access, this tool lets the AI request actions while the actual credential use and command execution happen on the server.
When automation workflows depend on API keys, tokens, or other sensitive data, this tool keeps secret management and execution on the server to reduce exposure risk.
It is an MCP server that lets AI agents use secrets stored in an encrypted vault, such as API keys and tokens, while executing HTTP, shell, and SSH actions on the server. Its key property is that secret values are never exposed to the AI.
From the description, it at least requires a server-side runtime and an encrypted vault for storing secrets. For exact installation and configuration requirements, see the source repository.
The difference is that secrets are used only on the server, while the AI can trigger actions and receive results without seeing the plaintext secret. That makes it better suited for automation involving protected resources.
Securely lets AI agents access and manage secrets through MCP tools.
Securely store and manage encrypted credentials for AI agents without hardcoded keys.
Securely store and use secrets in MCP clients without exposing them to models.
Build an API-key protected MCP server with realistic e-commerce tools and prompts.
Securely share and store secrets via claim codes without exposing them in chats.
Securely lets AI use real secrets with human approval per release.