feat: added user defined session_path and minor fixes

Author: STAR-173

README.md

@@ -1,27 +1,27 @@
 # LLMSession
 
 [![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](LICENSE)
-[![Build and Publish](https://github.com/star-173/llm_session/actions/workflows/publish.yml/badge.svg)](https://github.com/star-173/llm_session/actions/workflows/publish.yml)
 [![PyPI version](https://badge.fury.io/py/llm-session.svg)](https://pypi.org/project/llm-session/)
 [![Python 3.8+](https://img.shields.io/badge/python-3.8+-blue.svg)](https://www.python.org/downloads/)
 
-A zero-configuration tool to automate interactions with web-based LLM providers (currently ChatGPT). It handles authentication, session persistence, and chained prompt execution programmatically.
+A zero-configuration tool to automate interactions with web-based LLM providers (**ChatGPT** and **Claude**). It handles authentication, session persistence, and chained prompt execution programmatically.
 
 ## Features
+- **Multi-Provider Support**: Automate interactions with all ChatGPT, AIStudio and Claude (Anthropic).
 - **Zero-Config Setup**: Automatically handles browser binaries via Playwright.
 - **Session Persistence**: Reuses cookies/storage for subsequent runs (no need to login every time).
-- **Smart OTP Handling**: Supports non-blocking callbacks for 2FA/OTP challenges.
 - **Resilient**: Allows custom CSS selectors to adapt to UI changes without updating the library.
 
 ## Prerequisites
-You must set the following environment variables, or pass them directly to the constructor:
-- `CHATGPT_EMAIL`
-- `CHATGPT_PASSWORD`
-- `CHATGPT_GOOGLE_LOGIN` (Optional: set to "true" if using Google Auth. *Note: Google Auth is experimental and may require manual intervention.*)
+You must set the following environment variables, or pass them directly to the constructor via the `credentials` dictionary.(Currently, Use google account only in order to login)
+
+**Google Login:**
+- `LLM_EMAIL`
+- `LLM_PASSWORD`
 
 ## Disclaimer
 > [!WARNING]
-> **Cloudflare/Bot Detection**: Automated interactions with ChatGPT are subject to high-security bot detection (Cloudflare). This library uses standard browser automation and may be blocked. For production reliability, please use the Official OpenAI API.
+> **Cloudflare/Bot Detection**: Automated interactions with web providers are subject to high-security bot detection. This library uses standard browser automation and may be blocked. For production reliability, please use the Official OpenAI or Anthropic APIs.
 
 This tool automates a third-party web interface. It is subject to breakage if the target website changes its DOM structure. Use responsibly and in accordance with the provider's Terms of Service.
 
@@ -44,19 +44,18 @@ from llm_session import Automator
 # 1. Configure Standard Logging
 logging.basicConfig(level=logging.INFO)
 
-# 2. Define OTP Callback (Optional, but recommended for headless envs)
+# 2. Define OTP Callback (Optional)
 def my_otp_handler():
-    # In production, you might fetch this from an email API
     return input("Enter OTP Code sent to email: ")
 
-# 3. Initialize
+# 3. Initialize (Select 'chatgpt' or 'claude')
 bot = Automator(
-    provider="chatgpt",
-    headless=False,  # Set to True for production (CURRENTLY ONLY WORKS WITH headless=False)
+    provider="claude",  # Options: "chatgpt", "claude"
+    headless=False,     # Set to True for headless mode (may increase detection risk)
     credentials={
         "email": "your_email@example.com", 
         "password": "your_password",
-        "method": "email" # or "google"
+        "method": "google" # Claude provider defaults to Google Auth flows
     },
     on_otp_required=my_otp_handler
 )
@@ -81,7 +80,7 @@ bot.close()
 
 ### Custom Selectors
 
-Websites change their layout often. If ChatGPT updates their CSS class names, you don't need to wait for a package update. You can inject your own selectors during initialization.
+Websites change their layout often. If a provider updates their CSS class names, you don't need to wait for a package update. You can inject your own selectors during initialization.
 
 ```python
 bot = Automator(
@@ -103,25 +102,26 @@ Instead of passing credentials directly, you can use environment variables:
 ```python
 import os
 
-os.environ["CHATGPT_EMAIL"] = "your_email@example.com"
-os.environ["CHATGPT_PASSWORD"] = "your_password"
+os.environ["LLM_EMAIL"] = "your_email@example.com"
+# The library will detect these automatically
 
-bot = Automator(provider="chatgpt", headless=False)
+bot = Automator(provider="claude", headless=False)
 ```
 
 ---
 
 ## Session Management
 
-This library stores browser cookies and local storage in your OS's standard user data directory:
+This library stores browser cookies and local storage in your OS's standard user data directory. This allows the browser to maintain a "Logged In" state between script executions.
+
 - **Windows**: `%LOCALAPPDATA%\LLMSession`
 - **Linux**: `~/.local/share/LLMSession`
 - **macOS**: `~/Library/Application Support/LLMSession`
 
 **Key Features:**
-- **Persistence**: Sessions persist across reboots—no need to login every time
-- **Security**: Sensitive data is stored locally and never transmitted
-- **Reusable**: Once authenticated, subsequent runs will skip the login process
+- **Persistence**: Sessions persist across reboots.
+- **Context Isolation**: Each session runs in a persistent browser context.
+- **Security**: Sensitive data is stored locally and never transmitted.
 
 ---
 
@@ -145,15 +145,15 @@ Automator(
 ```
 
 **Parameters:**
-- `provider` (str): The LLM provider to use (currently only "chatgpt" is supported)
-- `headless` (bool): Whether to run browser in headless mode. Default: `False`
+- `provider` (str): The LLM provider to use. Supported: `"chatgpt"`, `"claude"`.
+- `headless` (bool): Whether to run browser in headless mode. Default: `False`.
 - `credentials` (dict): Dictionary containing login credentials:
-  - `email` (str): Login email
-  - `password` (str): Login password
-  - `method` (str): Authentication method - "email" or "google". Default: "email"
-- `session_path` (str, optional): Custom path for session storage. If not provided, uses OS default
-- `config` (dict, optional): Configuration options including custom selectors
-- `on_otp_required` (callable, optional): Callback function to handle OTP/2FA challenges
+  - `email` (str): Login email.
+  - `password` (str): Login password.
+  - `method` (str): "email" or "google".
+- `session_path` (str, optional): Custom path for session storage. If not provided, uses OS default.
+- `config` (dict, optional): Configuration options including custom selectors.
+- `on_otp_required` (callable, optional): Callback function to handle OTP/2FA challenges.
 
 #### Methods
 
@@ -188,102 +188,27 @@ bot.close()
 
 ---
 
-## Examples
-
-### Example 1: Basic Usage
-
-```python
-from llm_session import Automator
-
-bot = Automator(
-    provider="chatgpt",
-    headless=False,
-    credentials={
-        "email": "user@example.com",
-        "password": "password123"
-    }
-)
-
-response = bot.process_prompt("Explain quantum computing in simple terms.")
-print(response)
-
-bot.close()
-```
-
-### Example 2: Chained Prompts
-
-```python
-from llm_session import Automator
-
-bot = Automator(provider="chatgpt", headless=False)
-
-# Create a story step by step
-chain = [
-    "Write the first paragraph of a sci-fi story.",
-    "Continue this story: {{previous}}",
-    "Write a dramatic ending for: {{previous}}"
-]
-
-story_parts = bot.process_chain(chain)
-full_story = "\n\n".join(story_parts)
-print(full_story)
-
-bot.close()
-```
-
-### Example 3: With OTP Handler
-
-```python
-from llm_session import Automator
-
-def handle_otp():
-    """Fetch OTP from email or user input"""
-    code = input("Enter the OTP sent to your email: ")
-    return code
-
-bot = Automator(
-    provider="chatgpt",
-    headless=False,
-    credentials={
-        "email": "user@example.com",
-        "password": "password123"
-    },
-    on_otp_required=handle_otp
-)
-
-response = bot.process_prompt("Hello!")
-print(response)
-
-bot.close()
-```
-
----
-
 ## Troubleshooting
 
 ### Issue: Login fails with "Invalid credentials"
 **Solution**: 
-- Verify your email and password are correct
-- Check if you have 2FA enabled (provide `on_otp_required` callback)
-- Try logging in manually in a browser first to ensure your account is accessible
+- Verify your email and password.
+- Check if you have 2FA enabled (provide `on_otp_required` callback).
+- **Claude Users**: If using Google Auth, ensure the browser window (non-headless) allows you to click through any security prompts initially.
 
 ### Issue: "Cloudflare challenge detected"
 **Solution**: 
-- This library uses standard browser automation which may be detected
-- Try running with `headless=False` to solve CAPTCHA manually
-- For production use, consider using the official OpenAI API instead
+- This library uses standard browser automation which may be detected.
+- Try running with `headless=False` to solve CAPTCHA manually.
 
-### Issue: Selectors not working after ChatGPT update
-**Solution**: 
-- Use the custom selectors feature to update CSS selectors
-- Or wait for a library update with fixed selectors
-- Check GitHub issues for community-reported solutions
+### Issue: Popup not closing (Claude/Google Auth)
+**Solution**:
+- The library attempts to handle Google's "Continue" interstitial screens. If it gets stuck, manual intervention in `headless=False` mode usually fixes the session for future headless runs.
 
 ### Issue: Session not persisting
 **Solution**: 
-- Ensure the session directory has write permissions
-- Check if antivirus is blocking file writes
-- Try specifying a custom `session_path` with known write access
+- Ensure the session directory has write permissions.
+- Check if antivirus is blocking file writes.
 
 ---
 
@@ -299,12 +224,4 @@ We welcome contributions! Please see [CONTRIBUTING.md](CONTRIBUTING.md) for guid
 
 ## License
 
-This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details.
-
----
-
-## Support
-
-- **Issues**: [GitHub Issues](https://github.com/star-173/llm_session/issues)
-- **Discussions**: [GitHub Discussions](https://github.com/star-173/llm_session/discussions)
-
+This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details.

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "llm_session"
-version = "0.1.4"
+version = "0.1.5"
 description = "A zero-configuration tool to automate web-based LLM interactions."
 readme = "README.md"
 authors = [{ name = "LLM_Session", email = "shoaib.star.a@gmail.com" }]