prompt-architect MCP — Use Cases, Install & Live Demo

Why use it

Key features

7 frameworks: CO-STAR, RISEN, RISE, TIDD-EC, RTF, CoT, CoD
Picks the right framework for the task
Inline explanations so users learn the frameworks over time
Template export for reuse
Works as a prompt-authoring companion in Claude Code

Live Demo

What it looks like in practice

prompt-architect-skill.replay ▶ ready

0/0

Install

Pick your client

~/Library/Application Support/Claude/claude_desktop_config.json · Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "prompt-architect-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/ckelsoe/prompt-architect",
        "~/.claude/skills/prompt-architect"
      ],
      "_inferred": true
    }
  }
}

Open Claude Desktop → Settings → Developer → Edit Config. Restart after saving.

~/.cursor/mcp.json · .cursor/mcp.json

{
  "mcpServers": {
    "prompt-architect-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/ckelsoe/prompt-architect",
        "~/.claude/skills/prompt-architect"
      ],
      "_inferred": true
    }
  }
}

Cursor uses the same mcpServers schema as Claude Desktop. Project config wins over global.

VS Code → Cline → MCP Servers → Edit

{
  "mcpServers": {
    "prompt-architect-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/ckelsoe/prompt-architect",
        "~/.claude/skills/prompt-architect"
      ],
      "_inferred": true
    }
  }
}

Click the MCP Servers icon in the Cline sidebar, then "Edit Configuration".

~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "prompt-architect-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/ckelsoe/prompt-architect",
        "~/.claude/skills/prompt-architect"
      ],
      "_inferred": true
    }
  }
}

Same shape as Claude Desktop. Restart Windsurf to pick up changes.

~/.continue/config.json

{
  "mcpServers": [
    {
      "name": "prompt-architect-skill",
      "command": "git",
      "args": [
        "clone",
        "https://github.com/ckelsoe/prompt-architect",
        "~/.claude/skills/prompt-architect"
      ]
    }
  ]
}

Continue uses an array of server objects rather than a map.

~/.config/zed/settings.json

{
  "context_servers": {
    "prompt-architect-skill": {
      "command": {
        "path": "git",
        "args": [
          "clone",
          "https://github.com/ckelsoe/prompt-architect",
          "~/.claude/skills/prompt-architect"
        ]
      }
    }
  }
}

Add to context_servers. Zed hot-reloads on save.

claude mcp add prompt-architect-skill -- git clone https://github.com/ckelsoe/prompt-architect ~/.claude/skills/prompt-architect

One-liner. Verify with claude mcp list. Remove with claude mcp remove.

Use Cases

Real-world ways to use prompt-architect

How to turn a vague prompt into a structured one

👤 Anyone frustrated by inconsistent Claude output ⏱ ~10 min beginner

When to use: You keep asking Claude the same kind of question and getting different-quality answers.

Prerequisites

Skill installed — git clone https://github.com/ckelsoe/prompt-architect ~/.claude/skills/prompt-architect

Flow

Submit the vague prompt

Use prompt-architect. My current prompt: 'write something for our launch email'. Make this structured.✓ Copied

→ Framework recommendation + structured version
Pick + refine

Use CO-STAR. Tone: confident but human; audience: B2B SaaS founders; objective: drive webinar signups.✓ Copied

→ CO-STAR structured prompt with all fields filled
Save as template

Save this as template 'launch-email-costar'.✓ Copied

→ Template persisted

Outcome: A reusable structured prompt that produces consistent output.

Pitfalls

Framework overhead for simple tasks — For short, one-off tasks, Prompt Architect will recommend staying simple rather than forcing a framework

Combine with: claude-code-skill-factory-skill

Pick the right framework for a specific task shape

👤 Practitioners deciding CoT vs CO-STAR vs RISEN ⏱ ~15 min intermediate

When to use: You're not sure which framework fits.

Flow

Describe the task

Task: synthesize 20 customer interview notes into themes. Which framework, and why?✓ Copied

→ Framework recommendation with reasoning
Try an alternative

Now show the same task structured with CoT vs CO-STAR; compare.✓ Copied

→ Side-by-side with tradeoff notes

Outcome: Confident framework choice for the task shape.

Build a team-shared prompt pack

👤 Prompt engineers writing house prompts for a team ⏱ ~60 min intermediate

When to use: You want every teammate to get consistent Claude behavior on the same tasks.

Flow

Start a session

I need 8 team prompts: lead-qualification, onboarding-email, refund-decision, ... Walk me through each with the right framework.✓ Copied

→ Framework-per-task + drafts
Export as skill

Export the 8 prompts as a skill pack.✓ Copied

→ Skill directory ready to git clone share

Outcome: A versioned, shareable prompt pack.

Combine with: claude-code-skill-factory-skill

Combinations

Pair with other MCPs for X10 leverage

prompt-architect-skill + claude-code-skill-factory-skill

Architect prompts, then Factory packages them as a skill

Use prompt-architect for 5 prompts, then /build prompts with Factory to export them.✓ Copied

Tools

What this MCP exposes

Tool	Inputs	When to call	Cost
Transform prompt	vague prompt, framework?	Upgrading a prompt	Claude tokens
Recommend framework	task description	Unsure which to use	Claude tokens
Save template	name, prompt	When the prompt works well	0
Export pack	templates[]	Team sharing	0

Cost & Limits

What this costs to run

API quota: None
Tokens per call: 2-5k per prompt transformation
Monetary: Free
Tip: Once a template works, reuse it — don't re-architect every time.

Security

Permissions, secrets, blast radius

Credential storage: No credentials

Data egress: Prompt drafts go to Claude

Troubleshooting

Common errors and fixes

Framework feels like overkill for simple tasks

Prompt Architect should recommend minimal structure; if it doesn't, explicitly request 'minimal structure'

Output more verbose than original prompt

Structured prompts can be longer but should produce more consistent output; measure consistency rather than length

Alternatives

prompt-architect vs others

Alternative	When to use it instead	Tradeoff
Anthropic's prompt generator	You want Anthropic's own suggestions for Claude prompts	Fewer frameworks surfaced by name

More

Resources

📖 Read the official README on GitHub

🐙 Browse open issues

🔍 Browse all 400+ MCP servers and Skills