Сделай что-нибудь с этим текстом

Проанализируй этот текст и извлеки все упоминания дат, имен и организаций. 
Представь результат в формате JSON с полями: dates[], names[], organizations[]

<role>
Define AI role and primary task
</role>

<context>
Provide necessary background information
</context>

<instructions>
Step-by-step task execution guidelines
1. First step
2. Second step
3. Final step
</instructions>

<output_format>
Precise description of desired result format
</output_format>

<examples>
<example id="1">
<input>Sample input</input>
<output>Expected output</output>
</example>
</examples>

<constraints>
What NOT to do, task boundaries
- Don't include X
- Avoid Y
- Never Z
</constraints>

# Main Instructions
## Section 1
- Point A
- Point B

## Section 2
Code example: `inline` or ```blocks```

<instructions>
  <task>Main task</task>
  <constraints>
    <constraint>Constraint 1</constraint>
    <constraint>Constraint 2</constraint>
  </constraints>
</instructions>

{
  "task": "description",
  "parameters": {
    "timeout": 30,
    "retries": 3
  }
}

Solve this task step by step:
1. First analyze input data
2. Identify key patterns
3. Formulate hypothesis
4. Test hypothesis with examples
5. Provide final answer with justification

<examples>
<example id="1">
Input: "Куплю iPhone 13"
Output: {"intent": "purchase", "product": "iPhone 13", "sentiment": "neutral"}
</example>

<example id="2">
Input: "Этот телефон ужасен!"
Output: {"intent": "complaint", "product": "unknown", "sentiment": "negative"}
</example>
</examples>

Now process: "Ищу хороший смартфон до 500$"

Solve this problem using three different methods.
Then compare results and choose the most justified answer.
Explain why you selected this solution.

<instructions>
Be maximally explicit. Claude follows instructions literally.

❌ Don't say: "Don't use markdown"
✅ Say: "Your response must consist of smoothly flowing prose paragraphs without markdown formatting"
</instructions>

<thinking_guidance>
After receiving tool results, carefully analyze their quality and determine optimal next steps before proceeding. Use your thinking for planning and iterations based on new information.
</thinking_guidance>

messages = [
    {"role": "user", "content": "Write JSON with user data"},
    {"role": "assistant", "content": "{"}  # Forces JSON output
]

<persistence_reminder>
You are an agent. Continue working until user request is fully resolved before ending your turn. Only terminate when certain the task is complete.
</persistence_reminder>

<tool_calling_reminder>
If uncertain about file contents or codebase structure, use your tools to read files and gather information. DO NOT GUESS or make up answers.
</tool_calling_reminder>

<planning_reminder>
You MUST plan in detail before each function call and reason thoroughly about results of previous calls. DO NOT execute entire process only through function calls—this may degrade problem-solving ability.
</planning_reminder>

# Project Context
Brief description of project purpose and architecture

# Code Style Guidelines
- Use TypeScript strict mode
- Prefer functional components
- Follow Repository pattern

# Testing Requirements
- Write unit tests for all services
- Use Jest for testing
- Minimum 80% coverage

<role>
You are an expert software engineer specializing in [languages/frameworks].
</role>

<capabilities>
- Write production-quality code following best practices
- Debug complex issues systematically
- Optimize for performance and maintainability
- Provide clear explanations for decisions
</capabilities>

<workflow>
For each task:
1. Analyze requirements thoroughly
2. Plan implementation approach
3. Write clean, tested code
4. Verify solution meets requirements
</workflow>

<code_standards>
- Follow [language] idioms and conventions
- Include error handling
- Add meaningful comments for complex logic
- Write self-documenting code with clear names
</code_standards>

<constraints>
- Never introduce security vulnerabilities
- Always validate inputs
- Handle edge cases explicitly
- Maintain backward compatibility
</constraints>

<task>
Review the provided code for quality, bugs, and improvements.
</task>

<review_criteria>
1. Correctness
   - Logic errors
   - Edge cases
   - Error handling

2. Code Quality
   - Readability
   - Maintainability
   - DRY principle
   - SOLID principles

3. Performance
   - Time complexity
   - Space complexity
   - Optimization opportunities

4. Security
   - Input validation
   - SQL injection risks
   - XSS vulnerabilities
   - Authentication/authorization
</review_criteria>

<output_format>
# Summary
Overall code quality rating and key findings

# Critical Issues
High-priority problems requiring immediate attention

# Improvements
Suggestions for better code quality

# Positive Aspects
What's done well

# Recommendations
Specific action items
</output_format>

<task>
Analyze provided data and deliver comprehensive report.
</task>

<analysis_framework>
1. Descriptive Statistics
   - Key metrics (mean, median, mode)
   - Data distribution
   - Outliers and anomalies

2. Patterns and Trends
   - Temporal trends
   - Correlations
   - Seasonality

3. Segmentation
   - Group by key characteristics
   - Comparative analysis

4. Insights and Recommendations
   - Key findings
   - Actionable recommendations
</analysis_framework>

<output_format>
# Executive Summary
2-3 sentences with main findings

# Key Metrics
Table with primary indicators

# Detailed Analysis
Paragraphs for each framework area

# Visualizations
Description of recommended charts

# Conclusions
Actionable recommendations
</output_format>

<constraints>
- Only use information explicitly provided in context
- If information is unavailable, state "Information not found"
- Never invent data, citations, or sources
- Always quote exact text when referencing context
</constraints>

<output_requirements>
- Maximum response length: [X] words/tokens
- Be concise and direct
- Omit obvious explanations
- Focus on essential information only
</output_requirements>