Del archivo al resultado útil
Cómo pedir trabajo con esta skill instalada
Instalar la skill da un método a tu IA. Tu petición todavía debe aportar los hechos del caso, las restricciones y el entregable esperado.
Describe la decisión o el entregable, no solo el tema.
Añade materiales, público, límites y hechos conocidos.
Fija formato, criterios de calidad y comprobaciones.
Errores que conviene evitar
- Starting fieldwork before brief alignment: Rewrites cost 2x. Always do the alignment call first.
- Skipping quality review: Sending reports with wrong data destroys credibility. Never skip QA.
- Over-researching: 80% accuracy in 1 week beats 95% accuracy in 4 weeks. Define "good enough" upfront.
- Not flagging blockers early: Waiting until deadline to say "we're stuck" is fatal. Weekly updates surface issues in time.
- Insight debt: Not logging findings during fieldwork. Scrambling to reconstruct insights at the end is painful and error-prone.
- No clear "done" definition: Without acceptance criteria per task, scope creep is inevitable.