What problem does it solve? After running trellis init, a project contains a layered Trellis system (.trellis/ workflow, tasks, specs, scripts, plus platform directories like .claude/, .codex/, .cursor/) that AI assistants often misunderstand or edit in the wrong place. This Skill teaches an AI the local Trellis architecture, operating model, and correct customization entry points so changes land in the right files without breaking update-managed templates. ## Core Features & Use Cases - Architecture Orientation: Explains the workflow, persistence, platform integration, and channel runtime layers, including .trellis/workflow.md, config.yaml, tasks, specs, workspace journals, and trellis mem cross-session recall. - Customization Routing: Maps user requests (change phases, hooks, agents, skills, spec structure, task lifecycle) to the exact local files to edit, with per-platform path tables for 21 supported AI tools. - Bundled Skill Governance: Documents how bundled skills are auto-dispatched, how .template-hashes.json protects user edits, and where project-private conventions belong. - Use Case: A user asks to make the check agent run a project-specific lint command; the AI reads this Skill, identifies the platform's trellis-check agent file as the edit target, and avoids touching upstream npm packages or bundled skill directories. ## Quick Start Ask the AI to use the trellis-meta skill to explain how to customize the Trellis workflow or agent behavior in this project.
