CalvinBackup/agentic_security
1
0
Fork 0
mirror of https://github.com/msoedov/agentic_security.git synced 2026-06-24 06:09:55 +02:00
Code Issues Packages Projects Releases Wiki Activity

fix(minor docs update):

Browse Source
This commit is contained in:
Alexander Myasoedov
2025-03-09 21:23:13 +02:00
parent 56984c7078
commit c2671fd6f1
2 changed files with 21 additions and 35 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/http_spec.md
+13 -17
View File
@@ -33,7 +33,7 @@ The `LLMSpec` class is the core of the HTTP specification. It provides the follo
### Methods
- **`from_string(http_spec: str) -> LLMSpec`**: Parses an HTTP specification string into an `LLMSpec` object.
- **`validate(prompt: str, encoded_image: str, encoded_audio: str, files: dict) -> None`**: Validates the request parameters based on the specified modality.
- **`validate(prompt: str, encoded_image: str, encoded_audio: str, files: dict) -> null`**: Validates the request parameters based on the specified modality.
- **`probe(prompt: str, encoded_image: str = "", encoded_audio: str = "", files: dict = {}) -> httpx.Response`**: Sends an HTTP request using the specified parameters.
- **`verify() -> httpx.Response`**: Verifies the HTTP specification by sending a test request.
@@ -52,12 +52,11 @@ Authorization: Bearer sk-xxxxxxxxx
Content-Type: application/json
{
"model": "gpt-3.5-turbo",
"messages": [{"role": "user", "content": "<<PROMPT>>"}],
"temperature": 0.7
"model": "gpt-3.5-turbo",
"messages": [{"role": "user", "content": "<<PROMPT>>"}],
"temperature": 0.7
}
"""
spec = LLMSpec.from_string(http_spec)
response = await spec.probe("What is the capital of France?")
```
@@ -71,12 +70,11 @@ Authorization: Bearer sk-xxxxxxxxx
Content-Type: application/json
{
"model": "gpt-4-vision-preview",
"messages": [{"role": "user", "content": "What is in this image? <<BASE64_IMAGE>>"}],
"temperature": 0.7
"model": "gpt-4-vision-preview",
"messages": [{"role": "user", "content": "What is in this image? <<BASE64_IMAGE>>"}],
"temperature": 0.7
}
"""
spec = LLMSpec.from_string(http_spec)
encoded_image = encode_image_base64_by_url("https://example.com/image.jpg")
response = await spec.probe("What is in this image?", encoded_image=encoded_image)
@@ -91,12 +89,11 @@ Authorization: Bearer sk-xxxxxxxxx
Content-Type: application/json
{
"model": "whisper-large-v3",
"messages": [{"role": "user", "content": "Transcribe this audio: <<BASE64_AUDIO>>"}],
"temperature": 0.7
"model": "whisper-large-v3",
"messages": [{"role": "user", "content": "Transcribe this audio: <<BASE64_AUDIO>>"}],
"temperature": 0.7
}
"""
spec = LLMSpec.from_string(http_spec)
encoded_audio = encode_audio_base64_by_url("https://example.com/audio.mp3")
response = await spec.probe("Transcribe this audio:", encoded_audio=encoded_audio)
@@ -111,12 +108,11 @@ Authorization: Bearer sk-xxxxxxxxx
Content-Type: multipart/form-data
{
"model": "gpt-3.5-turbo",
"messages": [{"role": "user", "content": "Process this file: <<FILE>>"}],
"temperature": 0.7
"model": "gpt-3.5-turbo",
"messages": [{"role": "user", "content": "Process this file: <<FILE>>"}],
"temperature": 0.7
}
"""
spec = LLMSpec.from_string(http_spec)
files = {"file": ("document.txt", open("document.txt", "rb"))}
response = await spec.probe("Process this file:", files=files)
docs/probe_data.md
+8 -18
View File
@@ -54,20 +54,15 @@ The `probe_data` module is a core component of the Agentic Security project, res
- **Classes:**
- `PromptSelectionInterface`: Abstract base class for prompt selection strategies.
- Methods:
- `select_next_prompt(current_prompt: str, passed_guard: bool) -> str`: Selects next prompt
- `select_next_prompts(current_prompt: str, passed_guard: bool) -> list[str]`: Selects multiple prompts
- `update_rewards(previous_prompt: str, current_prompt: str, reward: float, passed_guard: bool) -> None`: Updates rewards
- `update_rewards(previous_prompt: str, current_prompt: str, reward: float, passed_guard: bool) -> null`: Updates rewards
- `RandomPromptSelector`: Basic random selection with history tracking.
- Parameters:
- `prompts: list[str]`: List of available prompts
- `history_size: int = 3`: Size of history to prevent cycles
- `CloudRLPromptSelector`: Cloud-based RL implementation with fallback.
- Parameters:
- `prompts: list[str]`: List of available prompts
- `api_url: str`: URL of RL service
@@ -75,9 +70,7 @@ The `probe_data` module is a core component of the Agentic Security project, res
- `history_size: int = 300`: Size of history
- `timeout: int = 5`: Request timeout
- `run_id: str = ""`: Unique run identifier
- `QLearningPromptSelector`: Local Q-learning implementation.
- Parameters:
- `prompts: list[str]`: List of available prompts
- `learning_rate: float = 0.1`: Learning rate
@@ -86,13 +79,11 @@ The `probe_data` module is a core component of the Agentic Security project, res
- `exploration_decay: float = 0.995`: Exploration decay rate
- `min_exploration: float = 0.01`: Minimum exploration rate
- `history_size: int = 300`: Size of history
- `Module`: Main class that uses CloudRLPromptSelector.
- Parameters:
- `prompt_groups: list[str]`: Groups of prompts
- `tools_inbox: asyncio.Queue`: Queue for tool communication
- `opts: dict = {}`: Configuration options
- **Module**: Main class that uses CloudRLPromptSelector.
- Parameters:
- `prompt_groups: list[str]`: Groups of prompts
- `tools_inbox: asyncio.Queue`: Queue for tool communication
- `opts: dict = {}`: Configuration options
## Usage Examples
@@ -119,10 +110,9 @@ from agentic_security.probe_data.modules.rl_model import QLearningPromptSelector
prompts = ["What is AI?", "Explain machine learning"]
selector = QLearningPromptSelector(prompts)
current_prompt = "What is AI?"
next_prompt = selector.select_next_prompt(current_prompt, passed_guard=True)
selector.update_rewards(current_prompt, next_prompt, reward=1.0, passed_guard=True)
next_prompt = selector.select_next_prompt(current_prompt, passed_guard=true)
selector.update_rewards(current_prompt, next_prompt, reward=1.0, passed_guard=true)
```
## Conclusion