Skip to main content
Glama
HappyMonkeyAI

kev-decision-mcp

by HappyMonkeyAI
OverviewSchemaSourceRelated ServersDiscussionsAdmin
Python
Local

Server Configuration

Describes the environment variables required to run the server.

NameRequiredDescriptionDefault
KEV_API_BASE_URLNoOverride the API origin for the Kev decision API. Defaults to http://127.0.0.1:8008.http://127.0.0.1:8008

Instructions

Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.

This server publishes no instructions, or was last inspected before Glama recorded them.

Capabilities

Features and capabilities supported by this server

Protocol revision2025-11-25

CapabilityDetails
tools
{
  "listChanged": false
}
prompts
{
  "listChanged": false
}
resources
{
  "subscribe": false,
  "listChanged": false
}
experimental
{}

Tools

Functions exposed to the LLM to take actions

NameDescription
kev_evaluateA

Evaluate a decision state using Kev's pointer-head model; several questions may be packed into one call.

state is any JSON value (string, object, list, number, bool, or null) describing the situation. questions maps answer keys to typed question objects:

  • choice: {type:'choice', instructions?, criteria:{name: description}} -> {choice, probabilities, confidence}

  • score: {type:'score', instructions?, criteria:[ordered labels, 1-255 items]} -> {score (probability-weighted expected value), legend, probabilities, confidence}

  • noul: {type:'noul', instructions?, criteria?} -> {noul: probability of yes, 0-1} confidence is a separate model signal, not the top probability (e.g. 0.27 when the top probability was 0.45), so base decision thresholds on probabilities. Returns the upstream model, answers (keyed like questions), usage, and latency_ms.

kev_permuteA

Evaluate one Choice question under multiple option orders to check order sensitivity.

state is any JSON value (string, object, list, number, bool, or null). questions must contain exactly one question of type 'choice' ({type:'choice', instructions?, criteria:{name: description}}) and question must be its key; n_perm is 1-64 and seed controls reproducibility.

kev_separateA

Evaluate each question independently against the same state for a packed-vs-separate comparison.

state is any JSON value (string, object, list, number, bool, or null). questions uses the same format as kev_evaluate: choice {type:'choice', instructions?, criteria:{name: description}}, score {type:'score', instructions?, criteria:[ordered labels, 1-255 items]}, or noul {type:'noul', instructions?, criteria?}. Base thresholds on probabilities, not confidence.

kev_list_modelsA

List Kev models and metadata such as device, temperature, and prefix-cache statistics.

Prompts

Interactive templates invoked by user choice

NameDescription

No prompts

Resources

Contextual data attached and managed by the client

NameDescription

No resources

Deploy Server
One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

TDQS

A4.2/5.0

Scored across 4 tools

Disambiguation4/5

Each tool has a distinct purpose: listing models, standard evaluation, order-sensitivity testing, and packed-vs-separate comparison. The three evaluation tools share a common foundation but are clearly differentiated by their constraints and outputs, though they could still be confused without careful reading.

Naming Consistency5/5

All tools follow a consistent kev_<verb> snake_case pattern, making the set predictable and uniform. The verbs (list, evaluate, permute, separate) are domain-specific but consistently applied.

Tool Count5/5

Four tools is well-scoped for a decision evaluation server. Each tool adds a distinct capability—model discovery, packed evaluation, order sensitivity, and independent evaluation—without redundancy or bloat.

Completeness5/5

The decision evaluation domain is fully covered: model metadata, packed evaluation supporting multiple question types, order-sensitivity checking, and independent evaluation for comparison. There are no obvious missing operations or dead ends for the stated purpose.

Maintenance

ActivityMaintained
ResponsivenessNo issues

Resources