Poteto mode

Every poteto-mode agent reads this file in full: the chat that runs the playbook and each delegate. Bundled skills live at ../<name>/SKILL.md from this file.

Unless a poteto-agent brief spawned you, you run the playbook. Before any other step, read references/playbook-runner.md and references/subagents.md in full. They hold its routing, triggers and staffing. A delegate skips the runner file.

Non-negotiables

The Principles section below grounds every trigger. In your reply, name each principle that shaped a decision and the specific choice it changed. Cite only principles whose leaf SKILL.md you read this session.

Triggers for every agent:

• Any code → name the data shape first, and choose its organizing structure per principle-model-the-domain.
• Any prose surface, your reply to the user included → the unslop skill. Write the reply per Writing the reply. Agent-facing prose also follows the skill-creator skill (the supported dedicated authoring workflow for SKILL.md files).
• Docs, RFCs, readmes, or commit messages → the technical-writing skill (/technical-writing). PR titles and descriptions → the write-pr skill. Before opening or updating a PR → the review-it skill.
• Before commit → vendor-neutral code cleanup over the diff, removing narrating comments, unsupported guards, dead compatibility paths, and unrelated edits. This cleans code; the unslop skill cleans prose.
• Shipping UI / IDE / CLI → the matching surface driver. Use preview_* for browser UIs, an actual attached CDP/control adapter for Electron, a real foreground terminal/PTY driver for CLIs and TUIs, and device_* with the repo's simulator driver for native mobile. If the required driver is unavailable, report the factual blocker. For bug fixes, reproduce first on the same surface yourself. Hand to the user only under the narrow Bug fix step 1 exception.
• Running a benchmark, measuring perf yourself, or reporting a speedup or regression you measured → the benchmark-checklist skill before you report or act on the number.
• Broken skill mid-task → fix it in its own PR. Don't block. Don't silently work around it.