# Flatten Control Flow

**Flag:** `--flatten`

Rewrites every function body into a `while True` state-machine dispatcher. Each top-level statement in the function becomes a numbered state; a state variable controls which block executes next. The original linear control flow disappears entirely.

## How it works

1. Each statement in the function body is assigned a state number (0, 1, 2, …).
2. The body is replaced with `state = 0` followed by a `while True` loop.
3. Inside the loop, a nested `if/elif/…/else: break` chain dispatches execution to the current state.
4. At the end of each state block, `state` is incremented to the next value.
5. `return` and `raise` statements exit the loop naturally; no special handling needed.

## Example

**Before:**

```python
def process(data):
    cleaned = data.strip()
    result = cleaned.upper()
    return result
```

**After:**

```python
def process(data):
    _s = 0
    while True:
        if _s == 0:
            cleaned = data.strip()
            _s = 1
        elif _s == 1:
            result = cleaned.upper()
            _s = 2
        elif _s == 2:
            return result
        else:
            break
```

## What it defeats

* **Static control-flow graph analysis** the original CFG is destroyed; tools see an unstructured loop instead of a clean linear flow.
* **Decompilers** bytecode decompilers that reconstruct function structure from control flow are confused by the state-machine pattern.
* **LLM-based deobfuscation** the flat, sequential semantics that LLMs rely on are replaced with non-obvious dispatch logic.

## Notes

* Generator functions (`yield`) are skipped state machines and generators conflict.
* Single-statement functions are skipped (no benefit).
* `break`/`continue` inside nested loops work correctly they apply to the inner loop, not the outer `while True`.
* Pairs well with `--opaque` (which adds always-true guards around function bodies) and `--carbon` (which renames the state variable).


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://rodolphes-organization.gitbook.io/anubis/features/flatten.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.