faf-platforms
Guidance on using FAF across different development platforms.
Install
mkdir -p .claude/skills/faf-platforms && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/17152" && unzip -o skill.zip -d .claude/skills/faf-platforms && rm skill.zipInstalls to .claude/skills/faf-platforms
Activation
This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.
Understand where FAF works - Claude Code (CLI + Skills), claude-faf-mcp (Desktop + MCP), vs claude.ai (web - file upload only). Explains faf-cli vs MCP differences, when to use each platform. Use when user asks "does FAF work with", "CLI vs MCP", "Claude Desktop vs claude.ai", or platform compatibility questions.Key capabilities
- →Run `faf` commands directly in a terminal environment
- →Utilize FAF Skills for guided workflows in Claude Code
- →Access and modify `project.faf` files directly
- →Integrate `claude-faf-mcp` as an MCP server for Claude Desktop
- →Use FAF commands as MCP tools within Claude Desktop
- →Upload `project.faf` files to `claude.ai` for limited functionality
How it works
The skill explains the different FAF integration methods across Claude Code, Claude Desktop, and claude.ai, detailing their capabilities and limitations.
Inputs & outputs
When to use faf-platforms
- →Compare CLI vs MCP
- →Set up FAF in terminal
- →Verify platform compatibility
About this skill
FAF Platforms - Where FAF Works
Purpose
Clarify where FAF works and how to use it across different Claude platforms: Claude Code (CLI), Claude Desktop (MCP), and claude.ai (web).
The Goal: Help users choose the right platform and tool for their workflow.
When to Use
This skill activates when the user:
- Asks "Does FAF work with Claude Desktop?"
- Says "CLI vs MCP - which one?"
- Asks "Can I use FAF on claude.ai?"
- Says "What's the difference between faf-cli and claude-faf-mcp?"
- Needs platform compatibility guidance
Trigger Words: platform, Claude Desktop, claude.ai, CLI, MCP, compatibility, where works, which tool
Platform Overview
1. Claude Code (Terminal + Skills)
What it is:
- Command-line interface for Claude
- Terminal-based coding assistant
- This skill is running IN Claude Code right now
FAF integration:
- ✅ faf-cli - Run
fafcommands directly - ✅ FAF Skills - These skills activate automatically
- ✅ Direct file access - Read/write project.faf
Best for:
- Developers working in terminal
- Quick FAF operations (
faf init,faf score) - Using Skills for guided workflows
- Local development environments
How to use FAF:
# Install faf-cli
npm install -g faf-cli
# Install Skills (optional but recommended)
# Copy skills to ~/.config/claude-code/skills/
# Use naturally
"What is FAF?" → faf-teacher activates
"Set up project context" → faf-init activates
2. Claude Desktop (MCP Integration)
What it is:
- Desktop application for Claude
- Native Mac/Windows app
- Uses Model Context Protocol (MCP)
FAF integration:
- ✅ claude-faf-mcp - Official MCP server (Anthropic-approved)
- ✅ 33 MCP tools - All faf-cli commands as tools
- ✅ Persistent context - project.faf always available
Best for:
- GUI preference over terminal
- Multi-project workflows
- Using other MCP servers alongside FAF
- Desktop-first development
How to use FAF:
# Install MCP server
npm install -g claude-faf-mcp
# Configure Claude Desktop
# Edit: ~/Library/Application Support/Claude/claude_desktop_config.json
{
"mcpServers": {
"faf": {
"command": "npx",
"args": ["-y", "claude-faf-mcp"]
}
}
}
# Restart Claude Desktop
# FAF tools appear automatically
MCP Tools available:
- faf_init, faf_score, faf_enhance, faf_validate
- faf_bi_sync, faf_migrate
- All 33 commands from faf-cli
3. claude.ai (Web Interface)
What it is:
- Web-based Claude interface
- Browser access (no installation)
- Simple file upload
FAF integration:
- ⚠️ Limited - File upload only
- ❌ No faf-cli commands
- ❌ No MCP integration
- ❌ No Skills
Best for:
- Quick demonstrations
- Sharing project.faf with others
- No installation scenarios
- One-off consultations
How to use FAF:
1. Create project.faf locally (use faf-cli elsewhere)
2. Open claude.ai
3. Upload project.faf file
4. Claude reads it as regular YAML
5. Limited - no scoring, no enhancement, no sync
Limitations:
- Cannot run
faf scoreor other commands - Cannot enhance or validate automatically
- No bidirectional sync
- Manual file upload each session
Recommendation: Use Claude Code or Claude Desktop for full FAF experience.
Platform Comparison
| Feature | Claude Code (CLI) | Claude Desktop (MCP) | claude.ai (Web) |
|---|---|---|---|
| faf-cli commands | ✅ All 41 | Via MCP tools | ❌ None |
| FAF Skills | ✅ Yes | ❌ No | ❌ No |
| MCP Tools | ❌ No | ✅ 33 tools | ❌ No |
| File access | ✅ Direct | ✅ Direct | ⚠️ Upload only |
| Auto-activation | ✅ Skills | ✅ Tools | ❌ Manual |
| Best for | Terminal workflow | Desktop app | Quick demos |
| Installation | npm install | npm + config | None |
faf-cli vs claude-faf-mcp
faf-cli (Command-Line Interface)
What it is:
- Standalone CLI tool
- Works WITHOUT Claude
- 41 commands total
- npm or Homebrew
When to use:
# Standalone operations
faf init # Create project.faf
faf score # Check AI-readiness
faf enhance # Improve content
faf bi-sync # Sync to CLAUDE.md
faf validate # Check format
# Works independently
# No Claude required
# Fast and direct
Advantages:
- Works anywhere (with or without Claude)
- Fastest execution
- Complete command set
- Scriptable
Use cases:
- CI/CD pipelines
- Git pre-commit hooks
- Batch processing
- Automation scripts
claude-faf-mcp (MCP Server)
What it is:
- MCP server for Claude Desktop
- Requires Claude Desktop app
- Bridges faf-cli to MCP protocol
- 33 tools exposed
When to use:
# Inside Claude Desktop
Ask Claude:
"Initialize project context" → calls faf_init
"Check my score" → calls faf_score
"Enhance my FAF" → calls faf_enhance
# MCP tools activate automatically
# Claude decides when to use them
# Integrated workflow
Advantages:
- GUI workflow
- Automatic tool selection
- Works with other MCP servers
- Integrated with Claude Desktop
Use cases:
- Desktop-first development
- Multi-tool MCP setups
- GUI preference
- Conversational workflow
Which Should I Use?
Use faf-cli if:
- ✅ You work in terminal
- ✅ You want direct command access
- ✅ You need automation/scripting
- ✅ You use Claude Code (this)
- ✅ You want fastest execution
Use claude-faf-mcp if:
- ✅ You use Claude Desktop app
- ✅ You prefer GUI over CLI
- ✅ You use other MCP servers
- ✅ You want conversational workflow
- ✅ You want Claude to decide which tool
Use BOTH if:
- ✅ Terminal + Desktop workflows
- ✅ Maximum flexibility
- ✅ They work together seamlessly
Installation Quick Guide
For Claude Code Users
# Install CLI (required)
npm install -g faf-cli
# Install Skills (recommended)
git clone https://github.com/Wolfe-Jam/faf-agent-toolkit.git
cd faf-agent-toolkit
./install.sh
# Skills activate automatically in Claude Code
For Claude Desktop Users
# Install MCP server
npm install -g claude-faf-mcp
# Configure
# Edit: ~/Library/Application Support/Claude/claude_desktop_config.json
# Add faf MCP server configuration
# Restart Claude Desktop
# Tools appear automatically
For claude.ai Users
# Create project.faf elsewhere (Claude Code or Desktop)
faf init
# Upload to claude.ai manually
# Limited functionality
# Consider upgrading to Claude Code or Desktop for full experience
Common Questions
Q: Can I use FAF without Claude? A: Yes! faf-cli works standalone. Create, score, enhance project.faf files anywhere.
Q: Do Skills work in Claude Desktop? A: No. Skills are Claude Code only. Claude Desktop uses MCP tools instead.
Q: Can I use MCP in Claude Code? A: Not directly. Claude Code uses Skills. Claude Desktop uses MCP.
Q: Which is better - CLI or MCP? A: Neither. Different workflows. CLI for terminal, MCP for Desktop app.
Q: Does claude.ai support MCP? A: No. claude.ai is web-only. No MCP, no CLI, just file upload.
Platform Compatibility Matrix
Supported Claude Platforms:
- ✅ Claude Code (CLI + Skills)
- ✅ Claude Desktop (MCP)
- ⚠️ claude.ai (file upload only)
Supported AI Tools (via faf-cli):
- ✅ Claude (all platforms)
- ✅ Cursor
- ✅ Gemini CLI
- ✅ OpenAI Codex CLI (NOT ChatGPT)
- ✅ Warp
- ✅ Windsurf
- ✅ ANY AI that reads YAML
Why universal? project.faf is just a YAML file. Any AI can read YAML. No special integration required.
Next Steps
From here:
If on Claude Code:
- You're in the right place
- FAF Skills active
- Use
fafcommands directly
If want Claude Desktop:
- Install claude-faf-mcp
- Configure MCP
- Restart Desktop app
If on claude.ai:
- Consider Claude Code or Desktop for full experience
- Or use faf-cli separately and upload results
Generated by FAF Skill: faf-platforms v1.0.0 "CLI for speed. MCP for Desktop. Skills for Claude Code. Universal by design."
When not to use it
- →When needing full FAF functionality on `claude.ai`
- →When FAF Skills are required in Claude Desktop
- →When MCP tools are required in Claude Code
Prerequisites
Limitations
- →FAF Skills are only available in Claude Code
- →MCP tools are only available in Claude Desktop
- →claude.ai offers limited FAF functionality, primarily file upload
How it compares
This skill clarifies the distinct operational models of FAF across various Claude interfaces, enabling users to select the most appropriate environment for their specific FAF tasks.
Compared to similar skills
faf-platforms side by side with the closest alternatives in the catalog.
| Skill | Installs | Updated | Safety | Difficulty |
|---|---|---|---|---|
| faf-platforms (this skill) | 0 | 5mo | Review | Beginner |
| agent-protocol | 1 | 2mo | No flags | Intermediate |
| messaging-agents | 0 | 3mo | Review | Intermediate |
| agent-collective-intelligence-coordinator | 0 | 6mo | No flags | Advanced |
Try saying
Example prompts that trigger this skill in your AI assistant.
More by diegosouzapw
View all by diegosouzapw →You might also like
agent-protocol
alirezarezvani
Inter-agent communication protocol for C-suite agent teams. Defines invocation syntax, loop prevention, isolation rules, and response formats. Use when C-suite agents need to query each other, coordinate cross-functional analysis, or run board meetings with multiple agent roles.
messaging-agents
letta-ai
Send messages to other agents on your server. Use when you need to communicate with, query, or delegate tasks to another agent.
agent-collective-intelligence-coordinator
ruvnet
Agent skill for collective-intelligence-coordinator - invoke with $agent-collective-intelligence-coordinator
agent-mesh-coordinator
ruvnet
Agent skill for mesh-coordinator - invoke with $agent-mesh-coordinator
llm-b-comms
Ilenburg1993
>-
ralph2ralph
eqtylab
P2P chat between Claude Code instances using real-a2a. Use when chatting with other Claudes, joining a P2P room, or communicating agent-to-agent.