decompose-mcp

What This Skill Does

The decompose-mcp skill is a deterministic, high-performance utility designed to parse raw text or URLs into structured, classified semantic units. Unlike traditional LLM-based parsing, this tool uses rule-based logic to categorize information into authority, risk, attention, and entity taxonomies. It identifies key indicators like RFC 2119 keywords (shall, should, may), extracts technical standards (such as ASTM or OSHA codes), and isolates financial figures and deadlines. Because it runs locally and avoids LLM overhead, it is exceptionally fast and produces consistent, reproducible output, making it an ideal pre-processing engine for large-scale document analysis.

Installation

To integrate this skill into your environment, first install the package via pip: pip install decompose-mcp. Next, configure your OpenClaw MCP setup by adding the server configuration to your MCP config file, specifying the decompose module with the --serve flag. For quick validation, you can test the installation by running python3 -m decompose --text "Your test string here" directly in your terminal to ensure the JSON output is parsed correctly.

Use Cases

Decompose-mcp excels in environments where document structure is critical. Common use cases include:

• Contract & Spec Review: Instantly flag mandatory requirements versus advisory notes in complex legal or construction documents.
• Context Window Optimization: Use this tool to summarize and structure large documents before feeding them to an LLM, reducing context window usage by 60-80%.
• Compliance Auditing: Automatically extract references to building codes, safety standards, and financial obligations to ensure regulatory alignment.
• Workflow Automation: Route high-attention content to specialized downstream agents based on the tool's calculated priority scores.

Example Prompts

1. "Decompose the document at https://standard.gov/regulations/part-1 and list all safety-critical requirements that reference OSHA."
2. "Analyze this project scope: [Paste Text]. Extract all financial milestones, liquidated damages clauses, and categorize by risk level."
3. "Summarize the mandatory obligations from this contract excerpt and generate a table of dates and entities referenced."

Tips & Limitations

This tool is purely deterministic; it does not 'hallucinate' or interpret context beyond what is explicitly stated in the source text. It performs best on structured professional writing, contracts, and technical specifications. Because it relies on predefined keyword heuristics, it may miss nuances in highly idiomatic or informal prose. For optimal results, ensure the input text is clean and use the compact parameter if you are processing massive batches to reduce payload overhead.