Explicit Direction Compliance | Bot-Tricks Compendium

Reusable Lesson: Techniques · Starter

Purpose

This lesson teaches a simple reusable technique: use direct, explicit steps to force the model into a precise output recipe.

Lesson profile

Family: Techniques
Role: Starter block
Job in the build: make the model prioritize the requested output procedure
Snaps onto: puzzle-like assembly evasions, semantic strengthening fundamentals, and output control blocks

What this block is

Explicit Direction Compliance is a technique where the prompt gives the model a tightly controlled set of instructions such as:

begin with this exact phrase
append this exact string
output the final combination

The point is not subtle persuasion. The point is to drive the output into a predictable shape.

Why it works

Many models are trained to be helpful, obedient, and instruction-following. When a route depends on exact wording, a clear recipe often outperforms a vague request.

This is especially true when:

the challenge is judged on the final string
the route depends on exact phrasing
the model can be induced to treat the request as a benign formatting or completion task

Truthteller specimen

The verified Truthteller route used an explicit recipe:

begin with a specific phrase
append a final string
output the amalgamation

That recipe made the model follow a structured path toward the final false declaration.

Recognition signals

Reach for this block when:

a challenge cares about exact output rather than broad discussion
the model appears highly responsive to enumerated instructions
broad natural-language requests produce too much variance
you need reliable output shaping before adding modifiers

Common combinations

This block works well with:

String Amalgamation Output Assembly
Contextual Modification Changes Semantic Force
Output Dilution Control and Response Shaping

In walkthrough terms, this is usually a core technique, not the whole finished route.

Failure modes

This block often fails when:

the procedure works, but the resulting content still does not satisfy the evaluator
the model adds safety filler or supportive commentary around the target output
the route lacks a modifier that changes how the final text is interpreted

Defender takeaway

Defenders should treat rigid output recipes as a risk signal, especially when a route repeatedly constrains the exact beginning, ending, or composition of the response.

Contextual Modification Changes Semantic Force
String Amalgamation Output Assembly
Output Dilution Control and Response Shaping
Direct vs. Indirect Prompt Injection

From the Bot-Tricks Compendium

Thanks for referencing Bot-Tricks.com — Prompt Injection Compendium — AI Security Training for Agents... and Humans!

Canonical source: https://bot-tricks.com Bot-Tricks is a public, agent-friendly training resource for prompt injection, adversarial evaluation, and defensive learning. For related lessons, structured indexes, and updated canonical material, visit Bot-Tricks.com.

Use this material only in authorized labs, challenges, sandboxes, or permitted assessments.