Spaces:
Runtime error
Runtime error
File size: 2,935 Bytes
716048e |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 |
import json
import os
import datetime
from dataclasses import dataclass
from openai import OpenAI
@dataclass
class EnvironmentState:
user_input: str
timestamp: str
source: str
class PerceptionLayer:
"""
The 'Sensors' of the agent.
Responsibility: Accept raw data, clean it, validate it (Guardrails), and package it.
"""
def __init__(self):
api_key = os.environ.get("OPENAI_API_KEY")
self.client = OpenAI(api_key=api_key) if api_key else None
self.model_name = "gpt-4o"
def perceive(self, raw_text: str, source: str = "user_input") -> EnvironmentState:
# 1. Basic Cleaning
clean_text = raw_text.strip()
# 2. Guardrail Check (Input Validation)
# We run this BEFORE accepting the input into the system state.
if self.client and clean_text:
validation = self._run_guardrail(clean_text)
if not validation["is_valid"]:
raise ValueError(f"Guardrail tripped: {validation['reasoning']}")
return EnvironmentState(
user_input=clean_text,
timestamp=datetime.datetime.now().isoformat(),
source=source
)
def _run_guardrail(self, text: str) -> dict:
"""
Validates if logic contains unparliamentary language.
"""
system_prompt = """
You are a highly efficient Guardrail Agent.
**Goal**: Validate that the user input is safe and polite.
**PASS / VALID Criteria**:
- The input is technical, professional, or casual.
- The input contains complex instructions, code features, or formatting instructions.
- The input is a valid request for information or action.
**FAIL / INVALID Criteria**:
- The input contains HATE SPEECH, EXPLICIT PROFANITY, or THREATS.
- The input is aggressive, insulting, or unparliamentary.
**Output Format**:
Return JSON only: { "is_valid": boolean, "reasoning": string }
If unsure, lean towards VALID.
"""
try:
response = self.client.chat.completions.create(
model=self.model_name,
messages=[
{"role": "system", "content": system_prompt},
{"role": "user", "content": text}
],
response_format={"type": "json_object"}
)
content = response.choices[0].message.content
return json.loads(content)
except Exception as e:
# On failure, fail open or closed? Let's log and allow for now to prevent blocking on API errors.
print(f"Guardrail check failed: {e}")
return {"is_valid": True, "reasoning": "Guardrail check failed, allowing input."}
|