aitmpl-downloader/
├── SKILL.md                 # This file - overview and commands
├── scripts/
│   ├── download.py          # Download script (GitHub API)
│   └── update-index.py      # Generate fallback index
├── fallback-index.json      # Static index (API fallback)
└── categories/              # Category-specific guides
    ├── agents.md
    ├── commands.md
    ├── skills.md
    ├── mcps.md
    ├── settings.md
    ├── hooks.md
    └── plugins.md

Agent

Use category-downloader agent to download by category:

Task({
  subagent_type: "general-purpose",
  prompt: `Read .claude/agents/spec2impl/category-downloader.md and execute.
           Category: ${category}
           Requirements: ${requirements}`
})

Category	Output Directory	Guide
agents	`.claude/agents/`	categories/agents.md
commands	`.claude/commands/`	categories/commands.md
skills	`.claude/skills/`	categories/skills.md
mcps	`.mcp.json` (merge)	categories/mcps.md
settings	`.claude/settings.local.json`	categories/settings.md
hooks	`.claude/settings.local.json`	categories/hooks.md
plugins	Multiple locations	categories/plugins.md

Commands

# List all items in a category
python3 .claude/skills/spec2impl/aitmpl-downloader/scripts/download.py list --category agents --json

# Search across categories
python3 .claude/skills/spec2impl/aitmpl-downloader/scripts/download.py search "<query>" --json
python3 .claude/skills/spec2impl/aitmpl-downloader/scripts/download.py search "<query>" --category skills --json

# Download specific item
python3 .claude/skills/spec2impl/aitmpl-downloader/scripts/download.py get "<path>" --output .claude/agents

# View available categories
python3 .claude/skills/spec2impl/aitmpl-downloader/scripts/download.py categories

# Clear cache (get fresh data from GitHub)
python3 .claude/skills/spec2impl/aitmpl-downloader/scripts/download.py clear-cache

GitHub API Source

Base URL: https://api.github.com/repos/davila7/claude-code-templates/contents
Cache: 15 minutes (use --no-cache to bypass)
Rate Limit: Set GITHUB_TOKEN env var for higher limits (60 req/hour → 5000 req/hour)

Fallback System

GitHub API has rate limits (60 requests/hour without token). When the API fails, the downloader automatically uses a static fallback index.

How It Works

1. Check local cache (15 min TTL)
   ↓ (cache miss)
2. Fetch from GitHub API
   ↓ (rate limit hit / network error)
3. Use fallback-index.json (static backup)

Updating Fallback Index

When new templates are added to the upstream repository, update the fallback index:

# Requires GITHUB_TOKEN for higher rate limits
export GITHUB_TOKEN=your_token_here
python3 .claude/skills/spec2impl/aitmpl-downloader/scripts/update-index.py

This generates fallback-index.json with all available templates.

Rate Limit Status

When rate limited, you'll see:

Using fallback index for agents (API rate limit or network issue)

The fallback ensures the downloader always returns results, even when GitHub API is unavailable.

GitHub Paths

Category	GitHub Path
agents	`cli-tool/components/agents`
commands	`cli-tool/components/commands`
skills	`cli-tool/components/skills`
mcps	`cli-tool/components/mcps`
settings	`cli-tool/components/settings`
hooks	`cli-tool/components/hooks`
plugins	`.claude-plugin`

Output Format (JSON)

[
  {
    "category": "agents",
    "name": "frontend-developer",
    "description": "Agents from development-team",
    "path": "cli-tool/components/agents/development-team/frontend-developer.md",
    "subcategory": "development-team",
    "download_url": "https://raw.githubusercontent.com/..."
  }
]

Workflow

Read category guide - Understand available items and mapping
List/Search - Find matching templates
Download - Get files to local project
Verify - Confirm installation

s-hiraoku-aitmpl-downloader

aitmpl.com Template Downloader

Architecture