Esta página se muestra en inglés. Una traducción al español está en curso.
PromptsApr 9, 2026·3 min de lectura

Prompt Architect — 27 Frameworks for Expert Prompts

Transform vague prompts into structured, expert-level prompts using 27 research-backed frameworks across 7 intent categories. Works with Claude Code, ChatGPT, Cursor, and 30+ AI tools.

Introducción

Prompt Architect is a Claude Code skill that systematically transforms vague prompts into structured, expert-level prompts using 27 research-backed frameworks across 7 intent categories. With 96+ GitHub stars and MIT license, it scores every prompt on 5 dimensions (clarity, specificity, context, completeness, structure) and recommends the optimal framework for your use case. Works with Claude Code, ChatGPT, Gemini CLI, Cursor, GitHub Copilot, Windsurf, OpenAI Codex, and 30+ Agent Skills compatible tools. Setup time: under 2 minutes.

Best for: developers and prompt engineers who want consistent, high-quality prompts across AI coding tools. Works with: Claude Code, Cursor, ChatGPT, Gemini CLI, Windsurf, Copilot.


27 Research-Backed Prompt Frameworks

Prompt Architect organizes 27 frameworks into 7 intent categories, automatically recommending the best match for your task:

High Complexity Frameworks

  • CO-STAR — Best for content creation: Context, Objective, Style, Tone, Audience, Response
  • RISEN — Optimized for multi-step processes: Role, Instructions, Steps, End goal, Narrowing
  • CRISPE — For detailed creative work
  • BROKE — For constraint-heavy tasks

Medium Complexity Frameworks

  • TIDD-EC — Precision work balancing Task, Instructions, Do/Don't, Examples, Context
  • Chain of Thought — Step-by-step reasoning for logic tasks
  • Tree of Thought — Branching exploration for complex decisions
  • ReAct — Reasoning + Acting for agentic workflows
  • Chain of Density — Progressive summarization
  • Self-Refine — Iterative self-improvement
  • Plus 11 more: RISE-IE, RISE-IX, RACE, CARE, Skeleton of Thought, Step-Back, Least-to-Most, Plan-and-Solve (PS+), RCoT, RePEF, CAI Critique-Revise

Low Complexity Frameworks

  • RTF — Quick tasks: Role, Task, Format
  • BAB — Transformations: Before, After, Bridge
  • APE — Automated prompt engineering
  • CTF — Context, Task, Format
  • Plus: Devil's Advocate, Pre-Mortem, Reverse Role Prompting

How It Works — 5-Stage Pipeline

  1. Analysis — Evaluates your prompt across 5 dimensions (clarity, specificity, context, completeness, structure), each scored 1-10
  2. Recommendation — Proposes the optimal framework with reasoning specific to your use case
  3. Dialogue — Asks 3-5 targeted clarifying questions progressively
  4. Application — Applies the selected framework to transform your prompt
  5. Refinement — Iterates until satisfaction, with option to switch frameworks

Quality Scoring

Every prompt receives a composite score across:

Dimension What It Measures
Clarity Goal clarity, lack of ambiguity
Specificity Detail level in requirements
Context Background information provided
Completeness All necessary elements present
Structure Organization and formatting

FAQ

Q: What is Prompt Architect? A: A Claude Code skill that uses 27 research-backed frameworks to transform vague prompts into structured, expert-level prompts with quality scoring across 5 dimensions.

Q: Is Prompt Architect free? A: Yes, it is open source under the MIT license and free to use.

Q: How do I install Prompt Architect? A: Run npx @ckelsoe/prompt-architect — the interactive installer auto-detects your AI agents and guides the setup.


🙏

Fuente y agradecimientos

Created by ckelsoe. Licensed under MIT.

prompt-architect — ⭐ 96+

Thanks to ckelsoe for building a comprehensive prompt engineering toolkit that makes expert-level prompting accessible to all developers.

Discusión

Inicia sesión para unirte a la discusión.
Aún no hay comentarios. Sé el primero en compartir tus ideas.