Using projects.

A project in LADLE is a container for related chats that share configuration and knowledge. Create one from the sidebar (New Project). Chats inside a project inherit the project's custom instructions and see the project's knowledge files as context.

**When to use a project vs individual chats:** - Use a project when the SAME setup applies to many chats. Example: an ongoing book you're writing (your style guide + character list applies to every chat about it), a codebase you're refactoring (the module boundaries apply to every chat about it), a research topic (your reading list + framing applies). - Use individual chats for one-shot questions, unrelated exploration, or anything where the setup would just be overhead.

**Project instructions** are a shared system-prompt addition. Anything you'd otherwise re-paste at the top of every chat ("respond in the voice of X," "use the following style rules," "assume the reader already knows Y") goes here once. Applies to every chat in the project.

**Project knowledge files** are documents pinned to the project. Every chat in the project has them in context. Useful for reference docs the model should always know about — your style guide, an internal glossary, a spec you're building toward. Knowledge files count against the same context window as chat history; pin only what actually gets referenced.

**File-count limits:** projects have a soft limit around 30 knowledge files. Past that, load times slow and context bloats. If you're at the limit, split into two projects with narrower scope.

**Cross-project isolation:** projects don't see each other's chats or files. A chat in Project A is invisible to Project B. Memory (user-level facts) IS shared across projects.