On this page

Production infrastructure for AI agents. Self-hosted or cloud.

SystemPrompt is the secure harness where AI superagents execute. Claude Code, ChatGPT, Gemini—these are the brains. SystemPrompt is where they run safely, with real authentication, scoped permissions, and production-ready infrastructure.

The Harness Concept

Think of Kubernetes: it doesn't compete with your application—it's where your application runs. SystemPrompt doesn't compete with Claude Code—it's where Claude Code (and any MCP client) executes securely.

┌─────────────────────────────────────────────────────────────────┐
│                      YOUR SUPERAGENT                            │
│              (Claude Code, ChatGPT, Gemini)                     │
│                           │                                     │
│                           ▼                                     │
│  ┌───────────────────────────────────────────────────────────┐ │
│  │                    SYSTEMPROMPT                            │ │
│  │                                                            │ │
│  │  ┌──────────┐  ┌──────────┐  ┌──────────┐  ┌──────────┐  │ │
│  │  │  OAuth2  │  │   MCP    │  │  Agents  │  │Analytics │  │ │
│  │  │   Auth   │  │ Servers  │  │   A2A    │  │ & Logs   │  │ │
│  │  └──────────┘  └──────────┘  └──────────┘  └──────────┘  │ │
│  │                                                            │ │
│  │  ┌──────────┐  ┌──────────┐  ┌──────────┐  ┌──────────┐  │ │
│  │  │  Files   │  │ Content  │  │Scheduler │  │ Database │  │ │
│  │  │ Storage  │  │   CMS    │  │   Jobs   │  │ Postgres │  │ │
│  │  └──────────┘  └──────────┘  └──────────┘  └──────────┘  │ │
│  └───────────────────────────────────────────────────────────┘ │
│                           │                                     │
│                           ▼                                     │
│                     YOUR USERS                                  │
└─────────────────────────────────────────────────────────────────┘

Why Rust

SystemPrompt compiles to a 50MB single binary that deploys anywhere with zero runtime dependencies. No Python virtual environments. No Node modules. No Docker-in-Docker.

Property Benefit
No runtime Deploy to bare metal, Docker, or serverless
Memory-safe No buffer overflows or data races in multi-tenant environments
Async-first Handle thousands of concurrent connections on Tokio
Type-safe Catch errors at compile time, not in production 
# That's it. One file. Run anywhere.
./systemprompt infra services start --all

See systemprompt infra services --help for all options.

Open Standards

No vendor lock-in. Built on:

  • MCP — Model Context Protocol for tool and resource access
  • A2A — Agent-to-Agent protocol for multi-agent coordination
  • OAuth2/OIDC — Industry-standard authentication
  • WebAuthn — Passwordless authentication

Architecture Overview

33 crates organized into five layers. Dependencies flow downward only.

┌───────────────────────────────────────────────────────────────┐
│  ENTRY: api, cli                                              │
├───────────────────────────────────────────────────────────────┤
│  APP: runtime, scheduler, generator, sync                     │
├───────────────────────────────────────────────────────────────┤
│  DOMAIN: users, oauth, ai, agent, mcp, files, content         │
├───────────────────────────────────────────────────────────────┤
│  INFRA: database, events, security, config, logging           │
├───────────────────────────────────────────────────────────────┤
│  SHARED: models, traits, identifiers, extension               │
└───────────────────────────────────────────────────────────────┘

For detailed API documentation, see docs.rs/systemprompt.

What You Get

Capability Description
Authentication OAuth2/OIDC + WebAuthn for every operation
Multi-tenancy Per-user isolation with scoped permissions
MCP Hosting Production tool servers with OAuth
Agent Execution Sandboxed A2A protocol agents
File Storage Upload, serve, and permission files
Content Management Markdown publishing with search
Job Scheduling YAML cron with reliable execution
Audit Logging Complete operational audit trail

Next Steps

Previous Next
Documentation Agents