Avoid caching raw HTTP response objects in TMDb request helpers so Redis no longer fails on embedded locks. Add an isolated regression test for sync and async request snapshots.
Refs #5763
Read the latest LLM connection settings when building runtime clients so Web updates take effect immediately instead of reusing module-import defaults.
Closes#5757
Return configured download directories as API-ready save_path values so external integrations can choose download destinations without guessing local or remote path syntax.
Fixes#5737
Keep transfer recognition results ready for scraping without fetching images twice on the same path. Also ensure redo-by-path transfer recognition still populates image data before metadata scraping.
Default title and path recognition to skip image fetching, while keeping scrape entrypoints and transfer-to-scrape paths populated with image data. This preserves lightweight recognition behavior without breaking metadata scraping.
Route title and path lookups through the fallback-aware entrypoints so auxiliary matches can reuse pre-assist keywords without forcing image fetches in lightweight flows. Also reduce noisy agent shutdown logging during cleanup.
Add a global cleanup switch, per-table retention periods, and scheduler config reload support so data cleanup can be managed and applied without restarting.
Use FRONTEND_VERSION from version.py as the default frontend release target so local setup and auto-update install the matching frontend bundle.
Closes#5693
Allow PostgreSQL socket paths without forcing a TCP port and reuse a single URL builder for sync, async, and migration flows. Document Redis socket URLs and close the socket connection request. Closes#5720
Fetch all OpenList/AList fs/list pages when using the default per_page=0 to avoid truncating directories at 200 entries. Add an isolated regression test for the auto-pagination behavior.
Fixes#5723
Parent-path fallback should stop at shared download roots so an old root-level download cannot hijack unrelated manual transfers. Keep exact DownloadFiles matches allowed at the shared root to preserve valid no-subfolder lookups.
Closes#5716
Regression from afcd895f52
Follow-up to #5702
Support WebUI API Key auth, newer add responses, and cookie sync so qBittorrent 5.2 can connect reliably while keeping legacy fallback behavior.
Fixes#5724
Return only a 10KB preview to the agent so large command results do not overwhelm conversations while keeping the full output available for follow-up reads. Add pytest to the project dependencies to make the regression tests runnable in the project venv.
- Refactor MoviePilotToolSelectorMiddleware to perform tool selection once per agent execution and cache the result in state, avoiding redundant LLM calls for each model round.
- Add abefore_agent to select tools at the start of agent execution and store selected tool names in state.
- Update awrap_model_call to reuse cached tool selection from state for subsequent model calls.
- Enhance test coverage for tool selection caching and reuse logic.
- Improve error logging in skill version extraction.
- Add mechanism to always include core tools (e.g., file operations, command execution) in LLMToolSelectorMiddleware
- Update MoviePilotToolFactory to provide filtered always-include tool names based on loaded tools
- Set default LLM_MAX_TOOLS to 30 in config
- Refactor agent initialization to support always_include parameter
- Enhance tests to cover always_include logic and async agent creation
- Add tools for querying built-in and custom filter rules, and for adding, updating, and deleting custom rules and rule groups
- Refactor filter module to use shared builtin rule definitions
- Enhance rule group querying to include syntax guidance and usage references
- Add unittests for agent filter rule tools registration and parsing logic
- Add compatibility patch for langchain-openai responses API to ensure system messages are extracted as top-level instructions, addressing Codex endpoint requirements.
- Update provider list: add Alibaba, Volcengine, and Tencent TokenHub; adjust SiliconFlow and MiniMax endpoints; refine provider ordering and model list strategies.
- Extend models.dev-only listing logic for providers lacking stable models.list endpoints.
- Increase models.dev cache TTL for improved efficiency.
- Add tests for openai responses API and streaming compatibility patches.
- Move LLMHelper and related logic from app.helper.llm to app.agent.llm.helper
- Update all imports to reference new LLMHelper location
- Introduce app/agent/llm/__init__.py for internal LLM adapter exports
- Add llm.py API router with endpoints for model listing, provider auth, and test calls
- Remove legacy LLM endpoints from system.py
- Update requirements for langchain-anthropic and anthropic
- Refactor test_llm_helper_testcall.py for async LLMHelper usage and new import paths
jxxghp