bayestruth

davccavalcante/bayestruth
★ 1 stars TypeScript AI/LLM Updated today
Exact Bayesian trust and reputation scoring for tools, MCP servers, skills, and agents. Zero-dependency, TypeScript-first: closed-form Beta-Bernoulli posteriors, calibrated credible intervals, Thompson-sampling routing, time-decay, and a tamper-evident audit trail.
View on GitHub → Try with Claude — $10 free →

Quick Install

Copy the config for your editor. Some servers may need additional setup — check the README.

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "bayestruth": {
      "command": "npx",
      "args": [
        "-y",
        "davccavalcante/bayestruth"
      ]
    }
  }
}

README Excerpt

<p align="center"> <img src="https://raw.githubusercontent.com/davccavalcante/bayestruth/main/assets/bayestruth-diagram.svg" alt="How BayesTruth works: every tool, MCP server, or agent earns a calibrated trust score with a confidence range. vector-db scores 0.97 (trust), fetch-tool 0.78 (monitor), search-api dropped to 0.43 (distrust). Trust means use it, monitor means not enough evidence yet, distrust means route away." width="560">

Tools (2)

partitionweb_search

Topics

agent-reliabilitybayesianbeta-bernoullicalibrationcomplianceconjugate-priorcredible-intervalmcpmulti-armed-banditnodejsreputationthompson-samplingtrust-scoringtypescriptuncertainty-quantification