All Products
    MCP Plugin

    KeyRing

    Sovereign secrets for AI agents.

    Not just a plugin — the complete secret management product. KeyRing ships with the gateway, health and rotation tools, an encrypted guidebook, and an onboarding experience that gets your agents running securely in minutes.

    $9.99
    Free — Early Access
    Get KeyRing — $9.99 FreeSetup Guide

    Free during Early Access. Full product: gateway, tools, encrypted guidebook, and onboarding kit.

    How It Works

    Three steps from install to sovereign secret management.

    01

    Install the Gateway

    pip install "coreforged-mcp>=1.4.0" or use uvx. KeyRing ships pre-docked — no assembly required.

    02

    Ingest Your Secrets

    Drop secret files into the keyring directory. Run the ingest command. Done.

    03

    Agents Call Securely

    Your AI agent calls keyring_authenticated_request. KeyRing's tools never put an API key in context.

    What You Get

    KeyRing ships as a complete product across four layers, not just a plugin you wire up yourself.

    Kernel

    The Core

    CoreforgedMCP gateway with KeyRing pre-docked
    Encrypted secret store with Fernet at rest
    Scoped permissions per secret per tool
    Audit trail for every access
    Tools

    Beyond the Basics

    Health checker — verify all secrets are live
    Key rotation — update credentials without downtime
    Expiration tracker — warnings before keys expire
    Drift detector — catch config mismatches
    Intelligence

    The Guidebook

    Step-by-step setup guide (works without AI)
    Agent decision trees for common scenarios
    Encrypted AI-assisted troubleshooting (shard-gated)
    Per-tool deep-dive references
    Experience

    From Install to Running

    Onboarding project with first-run checklist
    Starter configs for common platforms
    Installable skills: Claude Code starts them by name, and any agent can read them as plain files
    Post-success menu with progressive tool discovery

    Built for Agent Security

    KeyRing's Tools Never Put an API Key in Context

    The encrypted store and its master key sit in ~/.coreforged/keyring/, and an agent that can read files there can decrypt the store without any KeyRing tool. Keep your agent and its file tools out of ~/.coreforged/.

    Scoped Permissions

    Each secret declares which tools can access it. An agent that needs Stripe can't touch your GitHub token.

    MCP-Native Architecture

    Runs as a plugin inside the CoreforgedMCP gateway. One server, full secret management — no extra processes.

    Zero Configuration

    Drop your secrets into the keyring directory, run the ingest command, and every agent has scoped access.

    Audit Trail

    Every secret access is logged: which agent, which service, when. Know exactly who touched what.

    Skeleton Discovery

    Agents read the skeleton file to understand what secrets exist and their types — never the actual values.

    Who Needs KeyRing?

    AI agents making authenticated API calls (Stripe, GitHub, Slack, GA4)
    Multi-agent systems where each agent needs different credentials
    Development environments where secrets must never leak into logs or context
    Agent sessions, in Claude Code or any other tool that runs MCP servers, that interact with external services
    Production MCP servers handling customer data
    $9.99Free

    Early Access

    Gateway + encrypted secret store
    Health, rotation, and expiration tools
    Encrypted guidebook with decision trees
    Onboarding kit with starter configs
    Unlimited agents and secrets
    Get KeyRing — $9.99 Free
    Secure checkout via Polar

    Questions

    Which AI assistants does it work with?

    Built with and for Claude, and tested on Claude Desktop and Claude Code. Created for any agent: the tools are a standard MCP server, and every section of the guide has a manual method. So any agent that can use MCP tools, like Codex or Zed, should be able to follow it.

    Open Gateway, Sovereign Secrets

    The CoreforgedMCP gateway is free and MIT-licensed. KeyRing adds the secret management layer that makes production agent deployments safe. Your keys, your infrastructure, your rules.