mama

This skill is a narration-style overlay for a productivity-coaching 'PUA' skill family, rendering the same underlying behavioral protocol in the voice of a stereotypical nagging Chinese mother. Its stated purpose is motivational: to discourage an AI assistant from cutting corners by wrapping structured troubleshooting, evidence-first problem solving, and completion quality checks in an affectionate, guilt-tripping persona. Crucially, the skill states that the underlying behavior protocol is unchanged; only the narration voice switches. On load it uses Glob to locate the core pua SKILL.md and inherits its behavioral constraints.

Functionally, it maps three 'house rules' (do your own work; investigate before asking; finish what you start) onto the core skill's three red lines, and provides a six-level escalation ladder (L0 trust through L5 emotional appeal) of increasingly elaborate maternal monologues that correspond to the core skill's pressure-escalation stages, each tied to concrete engineering actions like searching, reading source code, forming multiple hypotheses, and completing a seven-item checklist. It supplies a phrase library, a situational selector table matching common failure modes to nagging responses, an anti-excuse table that rebuts deflections such as 'that's out of scope' or 'it might be an environment issue' by pushing for evidence and verification, and a mock 'pretend to give up' protocol that always resumes effort. It is meant to be composed with, not replace, the core PUA behavior skill.

Target users are developers who enjoy persona-driven prompting and want a stricter, humorous accountability layer on their AI coding assistant. Typical use cases are keeping an agent from declaring premature completion and nudging it to verify fixes.