doge-shell-repo

name	doge-shell-repo
description	Use when working in the doge-shell repository or doge-shell リポジトリ. Routes to the right crate or narrower doge-shell skill, avoids broad reads, and chooses the smallest effective Rust validation command.

Doge Shell Repo

• Start with rg --files or rg -n; do not open broad files first.
• Read references/task-map.md first when the task type is already clear.
• Read references/package-map.md before choosing Cargo package names.
• Read references/read-boundaries.md before opening README.md or running broad tests.
• Read references/module-map.md only when ownership is unclear.
• Read references/test-scope.md before choosing cargo commands.
• Open README.md only for user-facing docs, config examples, or feature behavior that is described there.
• Prefer the smallest test or check that proves the change.
• Repo-local completion skill: doge-shell-repl-completion.
• Repo-local process skill: doge-shell-process-pty.
• Repo-local Lisp/config skill: doge-shell-lisp-config.
• Repo-local command palette skill: doge-shell-command-palette-ai.
• Repo-local builtin command skill: doge-shell-builtin-commands.
• Repo-local chat tool skill: doge-shell-chat-tools.
• Repo-local serve/web skill: doge-shell-serve-web.
• Repo-local notebook/markdown skill: doge-shell-notebook-markdown.
• Switch to a narrower skill when the task is clearly about completion, parser, process/PTY, prompt/terminal UI, environment/startup, Lisp/config, history/frecency, command palette/AI actions, builtin commands, serve/web, notebook/markdown, safety policy, chat tools, investigation, validation, or skill authoring.
• If a narrower skill is not installed in runtime, read its repo-local source at docs/ai/skills/<skill>/SKILL.md instead of installing every skill.