Refactor LLM Integration to Use securellm_adapter

[Vicbi]

Refactor LLM Integration to Use securellm_adapter

⚙️ Release Notes

• I've adapted the codebase to use securellm_adapter.py instead of direct OpenAI calls. Here's a summary of the changes:

  Files modified

  1. llm_query/securellm_adapter.py

  Added new functions and classes for compatibility:
  - llm_chat() — supports multi-turn conversations with system messages
  - query_llm() — convenience function for simple queries with system message
  - SecureLLMClient — drop-in replacement class for the OpenAI client pattern

  2. llm_query/LLM_analysis.py

  - Removed openai import, added securellm_adapter imports
  - query_openai() now wraps query_llm() from the adapter
  - api_key parameter is now optional (kept for backward compatibility)

  3. llm_query/ent_surgical_llm_analysis.py

  - Same changes as LLM_analysis.py
  - Updated all functions to use SecureLLM

  4. batch_query/batch_query.py

  - Removed openai import, added securellm_adapter imports
  - parallel_process_llm_cases() now uses query_llm() directly
  - api_key parameters made optional

  5. training/training_LLM.py

  - Updated imports to use securellm_adapter
  - ConversationalLLMAnalyzer class now uses llm_chat() instead of OpenAI client
  - All api_key parameters made optional

  6. pyproject.toml

  - Removed openai dependency (no longer directly used)
  - Added python-dotenv dependency (used by securellm_adapter)

  Usage

  Instead of passing an api_key, set the VAULT_SECRET_KEY environment variable:
 export VAULT_SECRET_KEY="your-vault-key"

  All existing function calls remain backward compatible — the api_key parameter is still accepted but ignored.
  
  

• I've created a CLI entrypoint for the ENT-LLM project.

  cli.py - CLI entrypoint with the following features:

  All the available models can be found in the securellm repo.
  Usage

  python cli.py --list-models

  python cli.py --model apim:claude-3.7

python cli.py --model apim:llama-3.3-70b --input cases.csv --output results.csv

python cli.py --model apim:gemini-2.5-pro-preview-05-06 --interactive

After installing the package (pip install -e .)

ent-llm --model apim:gpt-4.1 --input cases.csv

Options

  ┌───────────────┬───────┬─────────────────────────────────────────┐                                                                                             
  │     Flag      │ Short │               Description               │                                                                                             
  ├───────────────┼───────┼─────────────────────────────────────────┤                                                                                             
  │ --model       │ -m    │ Select LLM model                        │                                                                                             
  ├───────────────┼───────┼─────────────────────────────────────────┤                                                                                             
  │ --input       │ -i    │ Input CSV file path                     │                                                                                             
  ├───────────────┼───────┼─────────────────────────────────────────┤                                                                                             
  │ --output      │ -o    │ Output CSV file path                    │                                                                                             
  ├───────────────┼───────┼─────────────────────────────────────────┤                                                                                             
  │ --delay       │ -d    │ Delay between API calls (default: 0.2s) │                                                                                             
  ├───────────────┼───────┼─────────────────────────────────────────┤                                                                                             
  │ --interactive │ -I    │ Interactive query mode                  │                                                                                             
  ├───────────────┼───────┼─────────────────────────────────────────┤                                                                                             
  │ --list-models │ -l    │ List available models                   │                                                                                             
  ├───────────────┼───────┼─────────────────────────────────────────┤                                                                                             
  │ --verbose     │ -v    │ Enable verbose logging                  │                                                                                             
  └───────────────┴───────┴─────────────────────────────────────────┘                                      

  
  

📝 Code of Conduct & Contributing Guidelines

By submitting this pull request, you agree to follow our Coding Guidelines:

• I agree to follow the Coding Guidelines.