From upwork-scraper
Install all Upwork Scraper plugin dependencies automatically. Run this after installing the plugin from the marketplace or cloning the repo.
How this skill is triggered — by the user, by Claude, or both
Slash command
/upwork-scraper:setupThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Fully automated installation of everything needed to run the plugin. Run all steps sequentially — do NOT skip any step and do NOT ask for confirmation between steps.
Fully automated installation of everything needed to run the plugin. Run all steps sequentially — do NOT skip any step and do NOT ask for confirmation between steps.
This plugin uses an MCP server that provides tool_* functions for Upwork scraping. The MCP server auto-starts a Camoufox browser session for authentication and page scraping. All tools are prefixed mcp__upwork-scraper__* in Claude's tool list.
The MCP server will NOT connect until dependencies are installed. That's why this setup skill exists — it installs everything so the server can start on next launch.
After setup completes, the user MUST restart Claude Code. On restart, the MCP server connects automatically and all tool_* functions become available. The plugin works from any directory — the user does NOT need to be inside the plugin folder.
Set PLUGIN_ROOT for all subsequent commands. Try these in order until one works:
CLAUDE_PLUGIN_ROOT is set, use that../pyproject.toml exists and contains upwork-job-scraper, use $PWD.find ~/.claude/plugins/cache -maxdepth 4 -name "pyproject.toml" -path "*/upwork-scraper/*" 2>/dev/null | head -1
If found, use the directory containing that pyproject.toml./plugin install upwork-scraper@upwork-plugin-claude and try again." Then STOP.Store the resolved path and use it for ALL subsequent commands.
Run via Bash: uv --version
If uv is NOT found, install it automatically:
uname -s (returns "Linux", "Darwin", or contains "MINGW"/"MSYS" for Windows).powershell -Command "irm https://astral.sh/uv/install.ps1 | iex"
curl -LsSf https://astral.sh/uv/install.sh | sh
~/.local/bin/uv --version || uv --versionRun via Bash with timeout 120000:
uv sync --directory "$PLUGIN_ROOT"
If uv is at ~/.local/bin/uv (wasn't in PATH), use the full path.
Run via Bash with timeout 300000:
uv run --directory "$PLUGIN_ROOT" playwright install firefox
This downloads ~80MB. Be patient and use the 5-minute timeout.
Run via Bash:
test -f "$PLUGIN_ROOT/.env" || cp "$PLUGIN_ROOT/.env.example" "$PLUGIN_ROOT/.env"
Run via Bash:
uv run --directory "$PLUGIN_ROOT" python -c "from src.server import mcp; print('OK: MCP server ready')"
If all steps succeeded, print exactly this (filling in the actual values):
Installation complete!
Plugin root: <resolved path>
uv: <version>
Python packages: installed
Firefox browser: installed
Config: .env ready
IMPORTANT: You must restart Claude Code for the MCP server to connect.
Close this session and reopen Claude Code, then run:
/upwork-scraper:best-matches
If any step failed, show which step failed with the error output and suggest how to fix it. Do NOT print the success message if any step failed.
npx claudepluginhub southlab-ai/claude-plugin-marketplace --plugin upwork-scraperInstalls Firecrawl SDK and configures API key for web scraping in Node.js or Python projects. Verifies connection with code examples.
Customizes Cowork plugins for organization tools/workflows by replacing template placeholders or editing skills/connectors in desktop app.
Guides Claude Code plugin creation, structure (plugin.json, commands/agents/skills/hooks), installation (/plugin), marketplaces, team workflows, testing, debugging, and settings. Delegates to docs-management.