Infisical MCP Server
Infisical secrets management accessible via function calling for AI agents.
Data last scanned today · Reviewed today
Overview
This MCP server enables direct interaction with Infisical APIs to manage secrets, including creating, deleting, updating, and retrieving sensitive values. It is designed to work with various authentication methods such as Universal Auth and access tokens, making it compatible with both standard and self-hosted Infisical deployments.
Our verdict
With a security score of 40/100, this server is classified as high-risk. The primary concerns are that it is community-maintained rather than official and lacks an editorially reviewed authentication or read-only implementation. It should only be used by developers who are comfortable managing the security implications of exposing secret-management tools to an AI model.
- Supports both universal authentication and standard access tokens
- Contains only four direct dependencies to reduce supply-chain exposure
- Actively maintained with a commit in the last day
- Openly licensed under the Apache-2.0 standard
- Security architecture and read-only modes remain unreviewed
- Lacks verification from the official Infisical vendor
- No documentation regarding security audits or risk profiles
Setup
{
"mcpServers": {
"infisical-mcp-server": {
"args": [
"-y",
"@infisical/mcp"
],
"command": "npx"
}
}
}Paste into ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows), then restart Claude Desktop. Some servers require API keys or tokens as environment variables — check the project's README and add an "env" object if needed.
- No read-only mode — running this grants full read/write access.
- No built-in authentication — anyone with access to this config can invoke every tool it exposes.
Tools
| Tool | Description | Risk |
|---|---|---|
| create-secret | Create a new secret | medium |
| delete-secret | Delete a secret | high |
| update-secret | Update a secret | medium |
| list-secrets | Lists all secrets | low |
| get-secret | Get a single secret | low |
| create-project | Create a new project | medium |
| create-environment | Create a new environment | medium |
| create-folder | Create a new folder | medium |
| invite-members-to-project | Invite one or more members to a project | medium |
| list-projects | List all projects | low |
Compatibility
| Client | Local | Docker | Remote | Read-only |
|---|---|---|---|---|
| ChatGPT | ||||
| Claude Desktop | ||||
| Cursor | ||||
| VS Code | ||||
| Windsurf |
Frequently asked questions
›What authentication methods does the server support?
The server supports both universal-auth and access-token methods. You can specify the method using the INFISICAL_AUTH_METHOD environment variable, which defaults to universal-auth.
›Can I use this server with self-hosted Infisical instances?
Yes, you can configure a custom host URL by setting the INFISICAL_HOST_URL environment variable. It defaults to https://app.infisical.com if not specified.
›How can I restrict which tools the server exposes?
You can use the INFISICAL_ENABLED_TOOLS environment variable to provide a comma-separated allowlist of tools. Any tool not included in this list will be hidden and rejected if called.
›Is it possible to prevent secret values from appearing in the model's context?
Yes, by setting INFISICAL_MASK_SECRET_VALUES to true, secret values in tool responses will be replaced with <masked> while maintaining visibility of metadata like secret names and paths.
›How can I debug the server?
You can debug the server by building the project with npm run build and using the MCP Inspector tool to connect and test individual functions.
Changelog
Badge
Maintain this server? Add the live badge to your README.