Skip to main content
Glama

ARBITER

by fyi.grip
OverviewDiscussionsAdmin

arbiter_compare

Paid ARBITER finite-candidate measurement for contextual choice. Use when a caller already has an explicit candidate field and needs it ordered against a query, current state, intent, context, question, or perspective. query is the situation to measure. candidates is the finite caller-supplied field to order. top_k optionally limits returned results. Returns MCP text content containing the JSON result from POST /v1/compare with the supplied candidates ordered by measured coherence. Does not generate candidates or execute the selected action. Deterministic for identical inputs and engine state. Price: $0.01 USDC on Base per call. MCP identity: fyi.grip/arbiter.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
queryYesCurrent state, context, intent, question, or perspective to measure against the candidate field.
top_kNoOptional maximum number of highest-ranked candidates to return.
candidatesYesFinite caller-supplied candidate field to order. ARBITER measures these candidates and does not generate them.

Schema Changelog

Changes observed during successful MCP inspections.

  1. Changed4 schema fields changed
    • addedInput schema / properties / candidates / description
      Added value: +"Finite caller-supplied candidate field to order. ARBITER measures these candidates and does not generate them."
    • addedInput schema / properties / query / description
      Added value: +"Current state, context, intent, question, or perspective to measure against the candidate field."
    • addedInput schema / properties / top_k / description
      Added value: +"Optional maximum number of highest-ranked candidates to return."
    • removedInput schema / properties / use_freq
      Removed value: -{
-  "type": "boolean"
-}
  2. First observed

TDQS

A4.3/5.0
Behavior4/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

No annotations are present, so the description carries the full burden. It transparently discloses cost ($0.01 USDC), determinism, return type (MCP text content with JSON), and side-effect boundaries (does not generate or execute). This is strong behavioral disclosure, though it omits potential error/rate-limit details.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is compact yet information-dense, covering purpose, usage, parameters, output, cost, and identity in a logical flow. No sentence is filler; the only slight extra is the MCP identity line, which is minor.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness4/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given there is no output schema, the description explains the return format adequately. It also covers cost, determinism, and non-execution, making it complete enough for an agent to safely invoke the tool. Missing error scenarios are not critical for this simple tool.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 100%, so the schema already describes all parameters. The description adds useful semantic context by defining query as 'the situation to measure', candidates as 'finite caller-supplied field', and top_k as 'optionally limits returned results', which enriches the bare schema definitions.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly identifies the tool as measuring/ordering finite caller-supplied candidates against a query, using verbs like 'order' and 'measure'. It also distinguishes itself from generation by stating it does not generate candidates, which sets it apart from the sibling arbiter_embed.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description gives an explicit condition: 'Use when a caller already has an explicit candidate field and needs it ordered...' It also notes the tool does not generate candidates or execute actions, which helps the agent avoid misuse. However, it does not explicitly name the sibling tool or give a detailed when-not-to-use list, so it's slightly below perfect.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

All ARBITER toolsInspect
Try in Browser
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

Glama MCP Gateway

Add one secure layer between your agents and this server.

Resources