MCP For Claude Marketplace
Chain-of-Thought Skeleton Verifier

Chain-of-Thought Skeleton Verifier MCP Connector for Claude

A+

Validates the structural integrity and parsing patterns of ReAct and Chain-of-Thought agent outputs.

3 tools Official Updated Oct 1, 2026 Official Vinkius Partner
More Details Connect to Claude

The Chain-of-Thought Skeleton Verifier MCP server provides a specialized toolkit for auditing the 'anatomy' of AI reasoning processes. It allows developers to programmatically verify if agent outputs adhere to specific structural families, such as XML-style tags (e.g., ``) or keyword-based prefixes (e.g., Thought:). By using tools like verify_parsing_pattern, you can detect pattern mismatches, while check_structural_integrity identifies broken reasoning loops where actions are initiated without corresponding observations. Additionally, get_reasoning_stats provides quantitative metrics, including thought step counts and efficiency scores, to measure the reasoning density of your agentic workflows.

reactchain-of-thoughtparsingagent-auditingstructural-integrity

3 tools expose this connector's capabilities to your AI agent.

validate_sequence_flow

Checks if the order of identified blocks adheres to the logical ReAct loop

analyze_structure

Performs a deep structural scan of a raw agent output string to validate its skeleton

get_ratio_metrics

Calculates higher-level behavioral metrics derived from the structural analysis

See how to talk to your AI agent using Chain-of-Thought Skeleton Verifier.

Analyze this agent output: <thought>I need to check the weather.</thought><action>get_weather()</action>

The `check_structural_integrity` tool would flag this as having a broken loop because the action is not followed by an observation.

Does this text use XML tags or keyword prefixes? Text: Thought: I will search for the capital of France. Action: search('Paris')

The `verify_parsing_pattern` tool identifies this as using the 'KEYWORD_PREFIX' pattern.

Calculate the reasoning metrics for: Thought: Step 1. Action: Run. Observation: Done.

The `get_reasoning_stats` tool would report a thought step count of 1 and provide an efficiency score based on the action-to-thought ratio.

The `analyze_structure` tool performs a deep scan of raw agent text to verify if it follows tag-based or keyword-based patterns, checking for complete reasoning blocks.

Related Connectors

agent-output-deduplicator

agent-output-deduplicator MCP

3 tools Official

Detects and removes redundant outputs from multi-agent workflows using Jaccard similarity and n-gram overlap.

A+ View details →
Agent Prompt Versioning Engine

Agent Prompt Versioning Engine MCP

4 tools Official

Deterministic prompt version management and rollback calculator.

A+ View details →
Agent Fallback Chain Calculator

Agent Fallback Chain Calculator MCP

3 tools Official

Predict agent degradation paths, latency impacts, and success probabilities.

A+ View details →
Archimedes First Principles Prover

Archimedes First Principles Prover MCP

1 tools Official

An AI recommended restructuring 'because the industry leader does it that way.' That is analogy — not axiom. The foundational claim was never stated: at what volume does the current structure fail? The problem was never decomposed. The conclusion was asserted with 'obviously' — never derived. The boundaries were never tested: when does this approach cost more than the value it creates? Archimedes derived pi from polygon geometry — not from Egyptian estimates. This tool forces axiom examination, decomposition, proof derivation, boundary testing, and leverage point identification.

A+ View details →