echology-io/decompose

Overview

A Python library that splits text into classified semantic units such as authority, risk, attention, and entities. It operates deterministically without using a language model, making it predictable and lightweight for AI agent pipelines.

Best for

Best for
Developers building deterministic text decomposition components for AI agents without using LLMs

Use cases

• Extracting authoritativeness signals from text for fact-checking pipelines
• Identifying risk-related phrases in compliance or security documents
• Parsing attention-worthy segments for summarization or alerting

How to use

Install

pip install decompose-mcp

Tested with

Claude Code, Cursor, Windsurf

Example client config

{\n  "mcpServers": {\n    "decompose": {\n      "command": "uvx",\n      "args": ["decompose-mcp", "--serve"]\n    }\n  }\n}

Notes

A Python library that splits text into classified semantic units such as authority, risk, attention, and entities. It operates deterministically without using a language model, making it predictable and lightweight for AI agent pipelines.

9 stars on GitHub. Last updated 2026-05-24. Licensed MIT.

Use cases

• Extracting authoritativeness signals from text for fact-checking pipelines
• Identifying risk-related phrases in compliance or security documents
• Parsing attention-worthy segments for summarization or alerting

Pros

• Deterministic output with no reliance on LLMs, reducing cost and latency
• Lightweight and easy to integrate into existing Python projects
• Clearly defined output categories for straightforward downstream processing

Cons

• Limited to four predefined semantic categories, reducing flexibility for other use cases
• Very small community and low star count, indicating limited adoption and support
• May not capture nuanced or context-dependent meanings that an LLM could handle

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.