Input Validation — Animate | AI Wisdom

← Back to topic
🎞 ANIMATE · Mode 3 of 5 · +100 XP— LLM Input Validation
📋Schema
⚙️Validators
🔄Injection
✅Guard
🎯 Next: Test →
Step 1 / 4
Speed
Define Schema
ChatRequest Pydantic model defines field types and defaults💡 Analogy
Like a form with required fields — Pydantic enforces the contract before data enters your system.
Pydantic automatically validates types: user_id must be str, max_tokens must be int. Default 512 is applied if not provided.
Console Output
> ChatRequest schema: user_id(str), message(str), max_tokens(int=512)
●input_validation.py
Define SchemaL1–8
1from pydantic import BaseModel, field_validator, ValidationError
2import re
3 
4class ChatRequest(BaseModel):
5    user_id: str
6    message: str
7    max_tokens: int = 512
8 
9    @field_validator('message')
10    @classmethod
11    def validate_message(cls, v: str) -> str:
12        if len(v) > 4000:
13            raise ValueError("Message exceeds 4000 chars")
14        if len(v.strip()) == 0:
15            raise ValueError("Message cannot be empty")
16        return v.strip()
17 
18    @field_validator('max_tokens')
19    @classmethod
20    def validate_tokens(cls, v: int) -> int:
21        if not 1 <= v <= 4096:
22            raise ValueError("max_tokens must be 1-4096")
23        return v
24 
25INJECTION_PATTERNS = [
26    r"ignore previous instructions",
27    r"system prompt",
28    r"you are now",
29    r"disregard all",
30]
31 
32def detect_injection(text: str) -> bool:
33    text_lower = text.lower()
34    return any(re.search(p, text_lower) for p in INJECTION_PATTERNS)
35 
36def validate_input(data: dict) -> ChatRequest | None:
37    try:
38        req = ChatRequest(**data)
39        if detect_injection(req.message):
40            raise ValueError("Prompt injection detected")
41        return req
42    except ValidationError as e:
43        print(f"Validation error: {e}")
44        return None
44 lines
Timeline — 4 Steps
Define Schema
ChatRequest Pydantic model defines field types and defaults💡 Analogy
Like a form with required fields — Pydantic enforces the contract before data enters your system.
Pydantic automatically validates types: user_id must be str, max_tokens must be int. Default 512 is applied if not provided.
Console Output
> ChatRequest schema: user_id(str), message(str), max_tokens(int=512)

Was this page helpful?

Discussion

ChatRequest Pydantic model defines field types and defaults

💡 Analogy

Like a form with required fields — Pydantic enforces the contract before data enters your system.

Pydantic automatically validates types: user_id must be str, max_tokens must be int. Default 512 is applied if not provided.

Console Output

> ChatRequest schema: user_id(str), message(str), max_tokens(int=512)

ChatRequest Pydantic model defines field types and defaults

💡 Analogy

Like a form with required fields — Pydantic enforces the contract before data enters your system.

Pydantic automatically validates types: user_id must be str, max_tokens must be int. Default 512 is applied if not provided.

Console Output

> ChatRequest schema: user_id(str), message(str), max_tokens(int=512)