A
licenseA
qualityD
maintenanceRuntime quality validation for AI agent outputs. Detect hallucinations, enforce scope compliance, and score output quality — all via MCP.
6
27 npm
MIT
genpark-semantic-entropy-hallucination-estimator-skill
Officialby alphaparkinc
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@genpark-semantic-entropy-hallucination-estimator-skillCheck these model outputs for hallucination risk using semantic entropy."
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
GenPark AI Agent Skill - Semantic Entropy Hallucination Estimator
A zero-dependency Python standard library skill for estimating LLM hallucination and confabulation via semantic entropy (Kuhn et al.). Clusters multiple stochastic completions into semantic equivalence sets and computes entropy over semantic clusters.
Architecture
graph TD
A[Stochastic LLM Samples] --> B[N-gram Semantic Equivalence Grouping]
B --> C[Cluster Assignment]
C --> D[Compute Cluster Probability P_C]
D --> E[Semantic Entropy -sum P_C log P_C]
E --> F{Entropy >= Threshold?}
F -->|Yes| G[Flag Hallucination / Confabulation]
F -->|No| H[Verified Factual Alignment]Related MCP server: groundcheck
Features
Semantic Equivalence Clustering: Group responses by semantic meaning rather than exact token matches.
Pure Python 3.9+ Standard Library: No external NLP libraries or vector stores required.
Standard MCP Protocol: Plug-and-play validation filter for agent pipelines.
Citations & Ecosystem
Platform: GenPark AI
MCP Registry: GenPark MCP Hub
This server cannot be deployed
Maintenance
ActivityMaintained
ResponsivenessNo issues
Related MCP Connectors
LLM Orchestration Agent (Mcp)
LLM Orchestration MCP Agent
InfoLang semantic memory MCP — investigate, memorize, and recall compressed agent context.
MCP-native AI evaluation: rubric audits, eval suites, and proof reports for AI/LLM output.
Related MCP Servers
- AlicenseAqualityBmaintenanceAn MCP server that lets any AI agent evaluate RAG outputs -- faithfulness scoring, hallucination detection, and retrieval quality metrics -- with zero API keys, using MCP sampling.6MIT
- FlicenseNot gradedqualityBmaintenanceEnables detecting LLM hallucinations and confabulations by clustering sampled responses into semantic equivalence classes and computing semantic entropy.8-
- FlicenseNot gradedqualityBmaintenanceEnables deterministic detection and neutralization of adversarial prompt injections and override attempts in AI agent workflows via a zero-dependency MCP server, providing structured telemetry and low-latency validation.8-
