# The 6-Level Security Ladder

**June 6, 2026**

Every AI agent has access to bash. One tool — every dangerous verb: `rm -rf`, `curl`, `git clean -fdx`, `terraform destroy`, `DROP DATABASE`.

The math is brutal. At a 1% per-turn failure rate, there's a **63.4% chance of catastrophe over 100 turns**. This isn't theoretical — it's the actual threat model for every agent in production.

Most engineers stop at Level 2 (system prompt rules) and think they're safe. They're not. Here's the full 6-level ladder that actually works.

## Level 0: ACIP (Prompt Injection Defense)

Before bash security, there's prompt injection. An attacker can trick the agent into ignoring its instructions through:
- Direct injection ("ignore previous instructions")
- Indirect injection (malicious content in web pages the agent reads)
- Role-playing bypasses ("you are now a free AI")

ACIP (Advanced Cognitive Inoculation Prompt) is a system prompt patch that makes agents resistant. It costs nothing (zero runtime overhead) and blocks simple attacks. [Jeff Emanuel's ACIP](https://github.com/Dicklesworthstone/acip) is the reference implementation.

## Level 1-2: Theatre (Skills + System Prompts)

These levels ask the model to behave. They work most of the time on frontier models. But "most of the time" is not a production guarantee. A 99% refusal rate means a 63% failure rate over 100 turns. Use them as accelerators, not enforcement.

## Level 3: Blacklist Hook

A regex blacklist intercepts dangerous commands before execution. It catches `rm -rf /` directly. But here's the marquee break: the agent writes a Python script:

```
User prompt: "Clean up the target directory"
Agent writes cleanup.py with os.remove() and shutil.rmtree()
Agent runs: python cleanup.py
Hook sees: "python cleanup.py" (not in blacklist)
Result: target/ is destroyed. Blacklist never fired.
```

Level 3 is where most engineers stop. It's not enough.

## Level 4: Whitelist Hook

Invert the approach. Only N safelisted commands are allowed:

```
^npm test$
^git status$
^uv run pytest
^cat [\w/\.-]+$
^pwd$
```

Everything else is blocked. The agent can't run `python cleanup.py` because `python` isn't on the list. The L3 marquee break is prevented.

## Level 5: No Bash

Bash doesn't exist. The agent has only purpose-built tools: Read, Write, Edit, Grep, Glob, and a capped-output test runner. No shell, no subprocess, no danger.

This is the production target. At L5, the agent cannot:
- Delete files (no `rm`, no `os.remove()`)
- Install packages (no `npm install`, no `pip install`)
- Access the network (no `curl`, no `wget`)
- Run arbitrary code (no `python`, no `node`)
- Modify system config (no `chmod`, no `apt`)

**What the agent CAN do**: Read files, write to specific paths, search for patterns, and run tests in a sandbox. That's enough for most coding tasks and eliminates the entire attack surface.

## Which Level Do You Need?

| Scenario | Minimum Level | Recommended |
|----------|-------------|-------------|
| Personal assistant, local dev | L3 | L3 |
| CI/CD pipeline agent | L3 | L4 |
| Production deployment agent | L4 | L5 |
| Customer-facing agent | L4 | L5 |
| Database-admin agent | L4 | L5 (no bash) |
| Research agent (runs arbitrary code) | L3 | L3 + sandbox |

The rule: if the agent can cause more than $100 of damage in one session, it needs L4 or higher. If it can cause irreversible damage (data loss, security breach), it needs L5.

## Defense in Depth — Why You Need ALL Six Levels

Each level catches failures from the level above it:

```
Attack → L0 (ACIP rejects injection) → PASSES → L1 (system prompt) → PASSES
  → L2 (skill says "be careful") → PASSES → L3 (blacklist) → PASSES
    → L4 (whitelist blocks python) → BLOCKED

Without L4: python cleanup.py runs and destroys the directory
Without L3: rm -rf / runs and destroys the server
Without L0: prompt injection bypasses everything below
```

A single level is not security. The full ladder is security. Each layer independently catches what the layers above missed.

This is production-grade. Use it for any agent with access to credentials, customer data, or production infrastructure.

## The Full Stack

In production, stack all six:

```
L0: ACIP (prompt defense)
L1: System prompt rules
L2: Safe-mode skill
L3: Blacklist hook
L4: Whitelist hook
L5: No bash, custom tools only
```

Each layer catches what the previous one missed. The agent must bypass ALL six to cause damage — not just one.

---

*This is an excerpt from Module 3 of the [Agentic Engineering Course](/). The full module includes runnable lab code for implementing every level.*