v3.1.1

[3.1.1] - 202-01-20

Added

• Added new --provider flag to force openai or anthropic provider
• New --base-url flag for OpenAI compatible endpoints (eg. Ollama etc)
• env variable support: SKYLOS_LLM_PROVIDER, SKYLOS_LLM_BASE_URL, OPENAI_BASE_URL
• Auto API key bypass for local endpoints (localhost, 127.0.0.1, 0.0.0.0)
• Added agent for LLM assisted detection
• Added new cache and parallel processing functionalities
• More unit tests for LLM agents, cache and parallel processing

Fixed

• --gate flag now uploads scan results before exiting, enabling GitHub App check updates for Pro users
• Pre-commit hook now correctly returns exit code 1 when issues are found (use skylos . --gate)
• False positives for methods called via Protocol interface
• Fixed gatekeeper check_gate(..., strict=False) support so strict mode can be driven from config
• Fixed gatekeeper mismatch with tests
• Progress callback bug. Wrong variable name (progress vs progress_callback) fixed

Changed

• OpenAIAdapter now uses Chat Completions API (chat.completions.create) instead of Responses API
• Provider resolution now follows priority chain: CLI flag -> env variable -> model name inference