What is the Productivity Loop?

The Productivity Loop is a continuous improvement cycle where five specific personas (Auditor, Architect, Analyst, Refiner, and Compounder) work together to build on previous knowledge and compound improvements.

Can I use these personas individually?

Yes, you can apply any individual persona (like the Analyst or Architect) to a specific teammate in any custom agent blueprint to deepen their expertise and process.

How do I apply a persona to a teammate in Claude Code?

You can apply a persona by referencing its specific markdown file in the spawn prompt, which provides the agent with the necessary methodology and behavioral instructions.

Are these personas only for software engineering?

While several are optimized for engineering (like the Analyst), others like the Facilitator, Visionary, and Realist are designed for general productivity, strategy, and brainstorming tasks.

What is the benefit of using a persona over a simple role description?

Personas provide a structured methodology, specific scoring criteria, and defined interaction patterns, ensuring consistent, high-quality results that brief descriptions cannot achieve.

Agent Team Personas

Name: Agent Team Personas
Author: nthplusio

bynthplusio

0•

生产力与工作流

Transforms simple agent roles into methodology-driven teammates using deep behavioral profiles and structured interaction patterns.

This skill provides a library of eight specialized behavioral personas—including Architect, Analyst, and Refiner—that define how AI agents think, evaluate, and collaborate. By moving beyond simple role descriptions, these personas introduce rigorous methodologies, quantitative scoring criteria, and the 'Productivity Loop'—a continuous improvement cycle where agent outputs compound over time. It is an essential tool for developers and power users who need consistent, high-quality results from multi-agent systems in Claude Code.

主要功能

018 Specialized Personas: Includes Auditor, Architect, Analyst, Refiner, Compounder, Facilitator, Visionary, and Realist.

020 GitHub stars

03Quantitative Scoring: Agents use defined criteria to rate work quality (1-10), ensuring objective evaluation.

04Reference-Based Architecture: Inject deep behavioral contexts directly from standardized markdown reference files.

05Productivity Loop Framework: A built-in 5-stage cycle that ensures continuous improvement and knowledge accumulation.

06Structured Methodologies: Each persona follows named phases with specific required outputs and decision gates.

使用场景

01Standardizing technical reviews by applying the Analyst persona’s multi-pass methodology to code reviewers.

02Structuring complex brainstorming sessions using the Facilitator, Visionary, and Realist personas to manage divergence and convergence.

03Establishing a continuous software improvement cycle using the sequential Productivity Loop personas.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add nthplusio/functional-claude team-personas

For use in Claude.ai and ChatGPT

Download Skill