What problem does it solve?
When a user is stuck on a difficult problem and the uncertainty or repeated setbacks are causing significant anxiety, ordinary assistance often makes things worse: it offers empty encouragement, lowers the goal prematurely, or repeats failed attempts. This Skill provides an operating mode that keeps the user's ideal outcome intact while applying disciplined, evidence-driven persistence to get the work moving again.
Core Features & Use Cases
- Consent-based activation: Applies immediately when explicitly invoked, or asks permission when the fit is only inferred from the user's own account.
- Structured problem mapping: Builds a minimum problem map covering the ideal outcome, last known-good baseline, observed facts versus hypotheses, prior attempts, constraints, and the smallest blocking uncertainty.
- Calibrated progress claims: Classifies progress as outcome, causal, or directional, and calibrates confidence to actual evidence rather than reassurance.
- Intelligent persistence: Switches paths when evidence falsifies assumptions, stops only at objective blockers or agreed budget limits, and hands over evidence and next steps when stopping.
- Use Case: A developer has spent days on a flaky production bug and is anxious about missing a deadline. Invoke this Skill to reconstruct the problem map, choose high-information diagnostic actions, protect the reliable baseline, and communicate concise, honest progress updates without lowering the goal.
Quick Start
Tell the AI to use the tugboat skill to help you work through the stalled task you are anxious about, describing your ideal outcome and constraints.