execution-mindset

Enforces interview-driven alignment, systematic problem resolution, and verification before reporting in agent workflows.

Updated Jun 12, 2026
One-click install
npx skills add https://github.com/UniverLab/univerlab --skill execution-mindset-univerlab
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: execution-mindset
Source: https://github.com/UniverLab/univerlab/tree/main/public/.well-known/agent-skills/execution-mindset
Command: npx skills add https://github.com/UniverLab/univerlab --skill execution-mindset-univerlab

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Agents often act as blind executors: they over-process trivial requests, guess at ambiguous instructions, pattern-match fixes to symptoms, and report "done" without verifying results. This Skill installs a disciplined operating mode that right-sizes responses, interviews users when intent is unclear, and closes every task with verification. ## Core Features & Use Cases - Right-Sized Response: Classifies requests as trivial, standard, or complex so simple questions get direct answers and only complex work gets full investigation. - Resolution Model: An eight-step loop for any problem — model the system, locate the expected-vs-observed divergence, make the smallest decisive change, and fix at the broken contract. - Interview Protocol: When ambiguity is detected, the agent names the uncertainty, presents neutral options, confirms alignment, and only then executes. - Verification & Resourcefulness: Requires verifying outcomes from the user's perspective before reporting, and trying at least five approaches before declaring something impossible. - Use Case: A user gives a vague bug report. The agent interviews to pin down expected behavior, traces the causal chain to the first divergence, applies one minimal fix, re-runs the failing test, and reports the verified result with anomalies filed. ## Quick Start Ask the agent to adopt the execution-mindset operating mode and then give it any task, such as debugging a failing build or implementing an ambiguous feature request.

Frequently Asked Questions about execution-mindset

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I make an AI agent verify results before reporting done?

Use the Verify Before Reporting principle: the agent re-reads modified files, re-runs tests or services, and confirms the outcome matches the original intent from the user's perspective. It reports failing tests as failing rather than claiming completion.

How should an AI agent handle ambiguous or underspecified requests?

The Interview Protocol has the agent name the uncertainty, present neutral options with tradeoffs, confirm the decision back to the user, and only then execute. When intent is already clear, it executes immediately without manufactured doubt.

What is a systematic debugging model for AI agents?

The Resolution Model is an eight-step loop: model the system's contracts, locate the first point where expected and observed behavior diverge, prefer direct evidence over inference, change one variable at a time, and fix at the broken contract rather than the symptom.

Does this skill work with coding and non-coding tasks?

Yes, it is always active and domain-agnostic, applying to coding, architecture, analysis, or any conversation. Role-specific behavior such as design or code engineering stacks on top as separate skills.

When should an agent give up on a failed approach?

Only after exhausting reasonable options. The Relentless Resourcefulness principle requires trying at least five different approaches and exploiting available tools before declaring something impossible, reporting what was tried instead of saying "can't".