ptc-orchestration
Activate when user needs multi-URL scraping, browser automation pipelines, or efficient tool orchestration to reduce API round-trips and context usage.
When & Why to Use This Skill
The PTC Orchestration skill leverages Programmatic Tool Calling to streamline complex multi-tool workflows within the Claude ecosystem. By utilizing Python-based orchestration, it significantly optimizes performance by reducing API round-trips by 60% and context token usage by 65%. This skill is essential for developers looking to build high-efficiency agentic loops and sophisticated automation pipelines using MCP servers.
Use Cases
- Multi-URL Web Scraping: Efficiently fetch, compare, and summarize content from multiple web pages simultaneously to accelerate market research and data gathering.
- Complex Browser Automation: Chain sequential browser actions—such as navigation, element interaction, and data extraction—into a single optimized pipeline for seamless web task execution.
- High-Performance Tool Chaining: Orchestrate three or more sequential tool calls within a single execution block to minimize latency and maximize token efficiency in complex agentic workflows.
- MCP Server Integration: Seamlessly coordinate multiple Model Context Protocol (MCP) servers like FlareSolverr and BrowserMCP to handle advanced web-based challenges.
| name | ptc-orchestration |
|---|---|
| description | Activate when user needs multi-URL scraping, browser automation pipelines, or efficient tool orchestration to reduce API round-trips and context usage. |
PTC Orchestration Skill
Use Programmatic Tool Calling (PTC) for efficient multi-tool workflows. PTC allows Claude to write Python code that orchestrates multiple tool calls, reducing:
- API round-trips (60% fewer calls)
- Context token usage (65% reduction)
- Execution time (53% faster)
When to Use
Activate this skill when the user needs:
- Multi-URL scraping: Fetch and compare multiple web pages
- Browser automation: Chain multiple browser actions (navigate → snapshot → click → extract)
- Tool orchestration: Any workflow with 3+ sequential tool calls
Available Commands
CLI Usage
# Multi-URL scraping with summarization
python -m ptc_wrapper.cli scrape https://url1.com https://url2.com
# Browser automation pipeline
python -m ptc_wrapper.cli browser "Navigate to X, extract Y" --url https://start.com
# Custom PTC prompt
python -m ptc_wrapper.cli run "Your complex multi-tool task" --servers flaresolverr,browsermcp
# List available tools
python -m ptc_wrapper.cli list --tools
Python API
from ptc_wrapper import PTCClient
async with PTCClient() as client:
await client.load_mcp_servers(["flaresolverr"])
result = await client.scrape_urls(urls, summarize=True)
Setup
Ensure the wrapper is installed:
cd ~/.claude/tools/ptc-wrapper
uv pip install -e .
Key Features
- MCP Integration: Works with existing MCP servers (flaresolverr, browsermcp)
- Tool Search: Uses
allowed_callersto enable code execution - Input Examples: Auto-generates examples for better parameter accuracy
- Agentic Loop: Handles multi-turn tool execution automatically
Architecture
PTCClient → Anthropic API (with code_execution)
↓
MCPClient → MCP Servers (flaresolverr, browsermcp via stdio)
The wrapper adds:
code_execution_20250825tool to enable PTCallowed_callers: ["code_execution_20250825"]to each toolinput_examplesfor parameter accuracy- Beta header:
advanced-tool-use-2025-11-20