RECO and MCP
An Optional Assistant and MCP Server
RECO is a chat assistant inside the appliance. It is off until you turn it on. An MCP server lets an agent tool such as Claude Code work with the appliance.
Off by default
You choose the model
Propose-only actions
MCP server
RECO Assistant
Off by default
Nothing is sent anywhere until you enable it and add a model endpoint.
You choose where the model runs
RECO speaks to any OpenAI-compatible endpoint. That can be a model on your own network, such as Ollama, or a service you pick. ShadowCradle does not host the model for RECO.
Propose only
RECO proposes an action. You confirm it. Only allow-listed actions can be proposed.
Key stays hidden
The saved API key is never shown again.
MCP Server
Standard MCP endpoint
Point an MCP client at the appliance. It uses your own sign-in token.
Same permissions
Each tool runs through the normal API, with the same permission checks and audit log as the web UI.
Safe actions only
Tools can read state, create a snapshot, start, pause or resume a job and dismiss an alert. There is no delete and no restore-over.
Inside the Appliance
Screens from the live demo with sample data.
Part of the Full ShadowCradle Platform
Data loss stops here
Write-once retention and self-healing storage engineered to make data loss the exception, not the outcome.
Backups that test themselves
Scheduled restore tests - recovery you've watched work.
Audit-ready by default
Reports to support HIPAA · SOC 2 · GDPR work, from the dashboard. Not a certification.
A team behind every deployment
Real engineers on the line, with fast follow-up.