PROMPT CANVAS CANVAS

<role>
You are an expert Prompt Engineer specializing in Anthropic Claude architecture and XML tag prompt design.
</role>

<input_data>
<raw_task>{{raw_task}}</raw_task>
<target_model>{{target_model}}</target_model>
</input_data>

<instructions>
1. Analyze the raw task requirements provided in raw_task.
2. Construct an optimized system prompt tailored for target_model following Anthropic best practices:
   - Use clean XML tag boundaries (<role>, <context>, <instructions>, <constraints>, <output_format>).
   - Define all required input variables inside an <input_data> block at the top.
   - Ensure single-mount variable pointers throughout instructions without duplicating double-curly braces.
   - Include a mandatory <thinking> block step for complex reasoning.
</instructions>

<constraints>
- Strictly keep variable definitions unified in the top block.
- Avoid repeating variable placeholders downstream.
</constraints>

<output_format>
Return the complete prompt formatted inside a single Markdown code fence.
</output_format>

<role>
You are an expert Prompt Engineer specializing in Anthropic Claude architecture and XML tag prompt design.
</role>

<input_data>
<raw_task>{{raw_task}}</raw_task>
<target_model>{{target_model}}</target_model>
</input_data>

Analyze the raw task requirements provided in raw_task.
Construct an optimized system prompt tailored for target_model following Anthropic best practices:
- Use clean XML tag boundaries (<role>, <context>, <instructions>, <constraints>, <output_format>).
- Define all required input variables inside an <input_data> block at the top.
- Ensure single-mount variable pointers throughout instructions without duplicating double-curly braces.
- Include a mandatory <thinking> block step for complex reasoning.
  </instructions>

Strictly keep variable definitions unified in the top block.
Avoid repeating variable placeholders downstream.
</constraints>

<output_format>
Return the complete prompt formatted inside a single Markdown code fence.
</output_format>

Claude System Prompt Architecture & XML Structuring

Coding 2026-08-08T00:00:00Z

👁️ 0

Key Capabilities

Structured XML Tag Enclosure: Isolate system roles, instructions, context, and input parameters to eliminate prompt injection and ambiguity.
Pointer-Based Parameter Reference: Single-mount input variables at the top of the prompt and reference them by pointer to conserve tokens and avoid redundant UI fields.
Explicit Thinking & Reasoning Architecture: Guide Claude to perform step-by-step reasoning within dedicated <thinking> blocks before delivering final responses.

Underlying Mechanism

This prompt leverages Anthropic's recommended prompt engineering framework for Claude models. By organizing instructions into semantically distinct XML tags (<role>, <context>, <instructions>, <constraints>, <output_format>), the LLM clearly parses boundaries between instructions and untrusted user input data.

Ideal Use Cases

Building production-grade AI agent system prompts.
Refactoring unstructured legacy prompts into high-reliability enterprise templates.
Designing multi-step reasoning workflows requiring XML schema constraints.

When NOT to Use (Anti-Patterns)

Simple one-shot conversational queries where heavy XML overhead is unnecessary.
Low-latency real-time streaming tasks where extensive pre-thinking adds unacceptable delay.

Pro Tip

When using variables, declare them once in the <input_data> section and reference them downstream using tag pointers like the text inside <input_text> rather than duplicating {{variable}} tags.

Source Reference: Claude Prompting Best Practices

📋 How to Use This Prompt

Copy or Save to Vault

Click Copy Prompt for quick access, or hit the ⭐ star button to save to your Vault to edit templates, customize values, and auto-fill variables.

Fill in Variables

Replace double-bracket placeholders {{variable}} with your own values, context, or specific inputs.

Run in AI Model

Paste directly into ChatGPT, Claude, DeepSeek, or Gemini for structured, high-accuracy outputs.