rsdouglas/janee

Overview

rsdouglas/janee is a TypeScript library for managing secrets in AI agents using the Model Context Protocol (MCP). It provides a secure way to store and retrieve credentials such as API keys and tokens that agents can access at runtime.

Best for

Best for
Developers building MCP-compatible AI agents who need a straightforward way to handle secrets.

Use cases

• Store and retrieve API keys for MCP-based AI agents
• Inject sensitive credentials into agent prompts or tool calls
• Manage multiple secret values across different agent workflows

How to use

Install

npx @true-and-useful/create-gh-app create my-agent --owner @me

Tools exposed

• list_services
• manage_credential
• reload_config
• hmac-bybit
• hmac-okx
• hmac-mexc
• service-account
• github-app
• oauth1a-twitter
• aws-sigv4

Tested with

Claude Desktop, Claude Code, Cursor, ChatGPT

Notes

rsdouglas/janee is a TypeScript library for managing secrets in AI agents using the Model Context Protocol (MCP). It provides a secure way to store and retrieve credentials such as API keys and tokens that agents can access at runtime.

28 stars on GitHub. Last updated 2026-03-17. Licensed MIT.

Use cases

• Store and retrieve API keys for MCP-based AI agents
• Inject sensitive credentials into agent prompts or tool calls
• Manage multiple secret values across different agent workflows

Pros

• Lightweight and focused on AI agent security
• Open source with simple TypeScript API
• Directly integrates with MCP for context-aware secret access

Cons

• Low community adoption (28 GitHub stars) may indicate early stage
• Relies on MCP ecosystem, limiting use outside that protocol
• Not a full-featured vault; designed only for agent secret injection

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.