dogfood

Test task instructions by delegating execution to a contextless agent.

Updated Apr 4, 2026

Search Tags:#workflow #qa #delegation #agent #forensics #dogfood #instruction-testing

Ready to run

npx skills add https://github.com/nicsuzor/aops --skill dogfood-nicsuzor

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: dogfood
Source: https://github.com/nicsuzor/aops/tree/main/aops-claude/skills/dogfood
Command: npx skills add https://github.com/nicsuzor/aops --skill dogfood-nicsuzor

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Delegated instruction testing enables teams to assess whether a set of instructions can be executed by a contextless agent, capturing friction, iterating on guidance, and codifying learnings to improve future tasks.

Core Features & Use Cases

Outer-loop instruction drafting and evaluation: write instructions, commission contextless execution, observe outcomes, and tighten guidance.
Friction discovery and iterative refinement: identify where instructions fail, gather evidence, and refine requirements.
Documentation and governance for continuous improvement: codify best practices, metrics, and feedback loops to ensure repeatable quality.

Quick Start

Provide a contextless instruction that starts the dogfooding loop and records friction observed during execution.

Frequently Asked Questions about dogfood

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

FAQPage Schema

How do I test if task instructions can be executed by a contextless agent?▼

Test task instructions by delegating execution to a contextless agent, observing the results, and identifying friction. This process requires explicit data sources, structured evaluation criteria, and defined output formats to ensure safe, reproducible outcomes.

What is delegated instruction testing for workflow QA?▼

Delegated instruction testing is a workflow QA method that assesses whether instructions can be executed by a contextless agent. It captures friction, iterates on guidance, and codifies learnings to improve future task execution.

How do I discover and refine friction in AI agent instructions?▼

Discover friction in agent instructions by commissioning contextless execution and observing where tasks fail. Gather evidence from these failures and iteratively refine the requirements to tighten guidance for future runs.

Do I need explicit data sources to run instruction testing loops?▼

Yes, you need explicit data sources, structured evaluation criteria, and defined output formats to run instruction testing loops. These inputs ensure safe, reproducible results when delegating execution to a contextless agent.

What is the best way to codify learnings from evaluating instruction effectiveness?▼

Codify learnings from instruction evaluation by documenting best practices, metrics, and feedback loops across multi-phase testing cycles. This governance ensures repeatable quality and continuous improvement for delegated tasks.

Why does my contextless agent fail to follow complex task instructions?▼

A contextless agent fails to follow instructions when guidance lacks explicit data sources or defined output formats. Identify the exact friction points through observation, then refine the requirements to tighten the guidance.