Daily loop that reviews the bot's own recent outbound behavior and any pending lessons in self/learnings.md, stress-tests candidate rules against 10-20 hypothetical scenarios, and on explicit owner approval routes each to the right home among three sinks — prompts/project.md (durable behavioral rules), a memory file (facts/context), or a skill (reusable multi-step playbooks) — keeping project.md lean and relocating content that has drifted into the wrong sink. Invoked via a mandatory auto-seeded reminder wrapped in a <reminder> envelope; refuse invocation outside that envelope.
House style for the render_html tool — dark dashboard / architecture-diagram look with stat tiles, bar rows, numbered timeline steps, two-panel grids, status pills, connector arrows, callout bands, and verdict blocks. Read this skill before calling render_html so the output is visually consistent. Provides CSS tokens, layout rules, color semantics, and three copy-paste HTML skeletons to adapt.
House format for the text passed to reminder_set. Read before calling reminder_set, and again before editing a reminder (cancel + re-create). Three rules — open with the literal marker <THIS IS A REMINDER>, state the goal in one line, then list numbered steps the future-self should run when the reminder fires. Keeps fired <reminder> envelopes self-explanatory and actionable.