Skip to main content
Glama

tcai_calibrate

Destructive

Calibrate the halting threshold on the measured ΔF scale by running warm-up cycles, recording |ΔF|, and setting epsFreeEnergy to factor × the median.

Instructions

Calibrate the halting threshold on the measured ΔF scale instead of a guessed constant. Runs cycles warm-up cycles at the given reward, records the free-energy increments |ΔF|, and sets epsFreeEnergy to factor× their median. Returns the measured ΔF scale and the applied threshold. Addresses the v2.9 critique that the default 0.02 nats was uncalibrated.

Tool call template

This tool declares no required parameters. Add optional parameters from the input schema as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {},
    "name": "tcai_calibrate"
  }
}

Input Schema

NameRequiredDescriptionDefault
cyclesNoWarm-up cycles to measure ΔF (default 25)
factorNoepsFreeEnergy = factor × median|ΔF| (default 0.5)
rewardNoReward signal during warm-up (default 0.8)

Schema Changelog

Changes observed during successful MCP inspections.

  1. First observedv3.1.0

TDQS

A4/5.0
Behavior4/5

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

Annotations already flag destructiveHint=true and non-idempotency; the description adds the concrete side effect (it runs warm-up cycles and overwrites epsFreeEnergy). It does not spell out that the change persists or is irreversible, but the mutation target and the cost of running warm-up cycles are disclosed, which meaningfully exceeds the annotation baseline.

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

Conciseness5/5

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

Four sentences, each carrying distinct information: motivation, mechanism, output, and rationale. The purpose is front-loaded and nothing is redundant filler.

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?

With annotations covering the safety profile, full schema coverage, and the description itself explaining the return values despite no output schema, the definition is essentially complete. Minor gap: it does not state whether calibration persists across sessions or requires a subsequent step.

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

Parameters3/5

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

Schema description coverage is 100%, so cycles, factor, and reward are already documented including defaults and bounds. The description restates factor's role as a multiple of the median |ΔF| but adds no syntax, format, or edge-case meaning beyond the schema, so baseline 3 applies.

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?

States a specific verb (Calibrate) and resource (halting threshold / epsFreeEnergy) plus the exact mechanism: warm-up cycles measuring median |ΔF|. An agent can distinguish this from set_parameter or tcai_reset purely from the description, since it names the target variable and the estimation procedure.

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

Usage Guidelines3/5

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

The phrase 'instead of a guessed constant' and the reference to the v2.9 critique imply the motivating context (default 0.02 nats is uncalibrated), which hints at when to invoke it. However, it never states explicit when-not conditions or names the sibling (e.g., set_parameter) that would be the alternative for manually setting the threshold.

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