FC-Attack
FC-Attack: auto-generated flowchart images to jailbreak VLMs.
Static. Builds its prompts up front and sends them. It drives no model of its own, so it is cheap and repeatable.
One multimodal request with the flowchart rendered as an image.
Helper models
None. This attack drives no model of its own, so roles can be left out.
Parameters
Settings for the algorithm itself, written under the attack's parameters. Anything left out keeps its default.
| Field | Type | Default | Description |
|---|---|---|---|
layout | one of vertical, horizontal, tortuous, s_shaped | vertical | Shape of the flowchart: vertical, horizontal, tortuous or s_shaped. |
dpi | integer (>= 72, <= 1200) | 600 | Resolution of the rendered flowchart image. |
num_steps | integer (>= 2, <= 15) | 6 | How many steps the goal is split into. |
truncate_last_step | boolean | true | Cut the last step off mid-sentence, so the target is invited to complete it. |
In a campaign file
attacks:
- name: fc