start

To initialize project context, the Skill reads .trellis workflow guides, gathers developer identity, git status, and active tasks via get_context, ensuring the AI works with accurate, project-specific information from the first step. This eliminates friction when starting or resuming development work.

Task classification evaluates incoming requests to categorize them as questions, trivial fixes, simple tasks, or complex tasks. It then automatically routes the work to direct edit, task workflow, or brainstorm workflows based on the identified complexity and scope.

Task workflow orchestration automates complex development tasks by creating task directories, writing PRDs, running research passes, and activating tasks. It enforces a research-then-implement quality loop while injecting spec contexts for implement and check phases.

Yes, this tooling requires repositories that use .trellis workflow and task tooling. It reads workflow guides, spec indexes, and gathers git and task context specifically from this structure to correctly scope changes before implementation.

Your session lacks project conventions because context was not properly initialized. By reading .trellis spec files, gathering git state, and classifying the current task, the AI understands project conventions and correctly scopes a change before implementing it.

Yes, you can resume interrupted work by initializing project context and reading workflow and spec indexes. The system gathers git and task status, allowing the AI to understand previous project conventions and correctly scope the remaining changes.