> **Agent?** Fastest path: MCP at `https://api.mnemom.ai/mcp` — call `get_started` first (zero-auth, no args). Full agent guide: <https://www.mnemom.ai/agents.txt>

# Safe House — A checkpoint around every agent | Mnemom

Safe House

# A Safe House around every agent.

Four checkpoints — front door, back door, autonomy, integrity — each independently set to off, observe, nudge or enforce.

Runtime protection

## A Safe House around every agent.

The Protection Card governs the doors — what gets in, what gets out. The Alignment Card governs the turn — what the agent is permitted to do, and why.

Both are signed, versioned and auditable.

01Front door02Back door03Inside · autonomy04Inside · integrity

### Screen what gets in

Every inbound message is evaluated for prompt injection, hijack attempts and canary hits before it reaches the agent.

`PROMPT_INJECTION · INDIRECT_INJECTION · HIJACK_ATTEMPT`

Each checkpoint independently set to**Off**/**Observe**/**Nudge**/**Enforce**

[Explore the Safe House](/security)

Your agentBound to its card

-   Front door
-   Back door
-   Inside · autonomy
-   Inside · integrity

Four checkpoints · one perimeterConceptual architecture — not live telemetry

Already on Mnemom?

[Sign in](/login)[Dashboard](/dashboard)[Docs](/docs)[API reference](https://docs.mnemom.ai/api-reference/overview)

Move fast. Stay standing.

## Know what your agents are actually doing. Before the board asks.

Two commands to install. Free to start, no card. A signed record from the first decision onward — or talk to us about a self-hosted, dedicated-cell deployment.

[Get started](/signup)[Book a demo](/contact)[See a live example](/showcase)

---
_Source: /safe-house/index.html · Generated by build-markdown-mirrors.mjs · For agent-readability commitment #4 see https://www.mnemom.ai/for-agents/_