- Added support for skill metadata including mode and triggers in skill.js.
- Improved skill loading from a catalog file and enhanced error handling in command-loader.js.
- Introduced allowed paths in command policies to restrict command execution to specified directories.
- Updated command validation to check against allowed paths in command-policy.js.
- Enhanced configuration management to include allowed paths in config-store.js.
- Refactored tools to respect allowed paths during file operations in tools.js.
- Improved user feedback for command execution based on policy checks.

…sification

- Added a new module for handling project instructions, including loading and trimming content from AGENTS.md files.
- Introduced configuration options for enabling/disabling project instructions and setting character limits.
- Enhanced command classification in the shell to recognize various service patterns for frontend, backend, database, and Docker services.
- Implemented a bounded cache for ignore rules in project indexing to improve performance.
- Refactored language loading to use a bounded cache with error handling for better reliability.
- Updated chat runtime to include new configuration options for project instructions.
- Improved memory snapshot composition to optionally include project instructions in system prompts.

…cess