What problem does it solve?
Teams often confuse process completion with real delivery: polished weekly reports, changed metrics, and boss approval replace actual user-verified results. This Skill injects reminders drawn from the essays 《置身钉内》 and 《置身钉外》 to push work toward evidence chains, real user paths, and verifiable acceptance criteria.
Core Features & Use Cases
- Contextual Reminders: Outputs blockquote-style reminders using original essay memes (薛定谔的用户, 每日一包, 温室数据, 口径瑜伽) matched to scenarios like "the boss says it's fine" or "the weekly report looks great".
- Scenario Routing Table: Maps eight common workplace situations (report polishing, metric redefinition, self-declared completion, greenhouse beta data) to concrete corrective actions.
- Structured Execution Layer: For complex tasks, appends a goal/acceptance/action/evidence/status/risk block so reminders translate into trackable delivery steps.
- Default Flavor Setting: Persists the ding flavor to
~/.pua/config.json when the user asks to set it as default.
- Use Case: When a developer says "I'm done" without running tests, the Skill responds with a reminder that self-reported completion is only a candidate state, then directs them to run the build and paste the output as evidence.
Quick Start
Ask the agent to review my claim that this feature is finished and remind me what evidence I still need before reporting it as done.