Systems Intelligence Performative Commercial Benchmarking

create_experiment

Create controlled experiments for comparison studies by pairing a pilot configuration with a hypothesis and measurement plan. Enforces association-not-causation evidence standards, allowing causal claims only with proper design.

Instructions

Create an experiment configuration for controlled comparison studies. Experiments pair a pilot configuration with a hypothesis and measurement plan. Use create_pilot_configuration first to build the config, then pass it here. Experiments enforce the ASSOCIATION-not-CAUSATION evidence standard — controlled experiments may upgrade evidence to CAUSATION only with proper design. REQUIRES AUTHORIZATION — contact pilots@mos2es.org for pilot access. In the synthetic demo, this returns an authorization notice.

Input Schema

TableJSON Schema

Name	Required	Description	Default
`name`	Yes	Experiment name (e.g., 'Q3 Claude vs ChatGPT operator comparison', 'Context window expansion pilot — Team Alpha')
`configuration`	No	Pilot configuration object (JSON) — see list_pilot_options for available metrics, eval families, and benchmark classes. Can be generated by create_pilot_configuration.

Output Schema

TableJSON Schema

Name	Required	Description
`tool`	Yes	Tool name that was called
`error`	Yes	Error type if authorization fails (e.g., 'authorization_required')
`status`	No	Initial status of the experiment (e.g., 'draft', 'configured')
`message`	Yes	Human-readable status message
`created_at`	No	ISO timestamp of creation (on success)
`experiment_id`	No	ID of the created experiment (on success)

Schema Changelog

Changes observed during successful MCP inspections.

Changed8 schema fields changedv0.4.3Aug 28, 2026

changedInput schema / properties / configuration / description

Previous value: -"Pilot configuration object (JSON) — see list_pilot_options for available metrics, eval families, and benchmark classes"New value: +"Pilot configuration object (JSON) — see list_pilot_options for available metrics, eval families, and benchmark classes. Can be generated by create_pilot_configuration."

changedInput schema / properties / name / description

Previous value: -"Experiment name (e.g., 'Q3 Claude vs ChatGPT operator comparison')"New value: +"Experiment name (e.g., 'Q3 Claude vs ChatGPT operator comparison', 'Context window expansion pilot — Team Alpha')"

addedOutput schema / properties / created_at

Added value: +{
+  "description": "ISO timestamp of creation (on success)",
+  "type": "string"
+}

addedOutput schema / properties / error / description

Added value: +"Error type if authorization fails (e.g., 'authorization_required')"

addedOutput schema / properties / experiment_id

Added value: +{
+  "description": "ID of the created experiment (on success)",
+  "type": "string"
+}

addedOutput schema / properties / message / description
```
Added value: +"Human-readable status message"
```

addedOutput schema / properties / status

Added value: +{
+  "description": "Initial status of the experiment (e.g., 'draft', 'configured')",
+  "type": "string"
+}

addedOutput schema / properties / tool / description
```
Added value: +"Tool name that was called"
```

First observedv0.4.0Aug 27, 2026

TDQS

A4.1/5.0

Behavior4/5

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

Annotations indicate openWorldHint, non-idempotent, non-destructive. The description adds valuable behavioral context: it enforces the ASSOCIATION-not-CAUSATION evidence standard, requires authorization, and discloses that in the synthetic demo it returns an authorization notice. This goes beyond the annotations and helps set expectations, though it does not detail side effects or persistence.

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 moderately concise and well-structured. It opens with the core purpose, then explains the workflow and evidence standard, and ends with authorization and demo behavior. Every sentence adds new information, though the authorization sentence could arguably be more compact. Still, it is efficient and each sentence earns its place.

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?

The tool has an output schema and well-documented parameters. The description covers key context: the prerequisite workflow, the evidence standard, authorization, and demo behavior. It does not explicitly mention return values, but the output schema exists to cover that. The only minor gap is the lack of details about what happens after creation (e.g., storage or side effects), but given the available structured schema and annotations, this is sufficiently complete.

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?

The input schema already provides 100% description coverage for both parameters. The description adds a small extra nuance by explaining that the 'configuration' parameter can be generated by create_pilot_configuration, and references list_pilot_options for available metrics. This is helpful, but the schema already carries the main semantic load, so a baseline 3 is appropriate.

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 states the tool's purpose: 'Create an experiment configuration for controlled comparison studies.' It specifies the verb (create), the resource (experiment configuration), and the context (controlled comparison studies). It also distinguishes itself from the sibling create_pilot_configuration by explaining that experiments pair a pilot configuration with a hypothesis and measurement plan, and by explicitly instructing to use create_pilot_configuration first.

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 explicit usage context: 'Use create_pilot_configuration first to build the config, then pass it here.' It also notes authorization requirements and the synthetic demo behavior. While it does not explicitly say 'when not to use,' it provides a clear workflow that guides an agent to the correct alternative when appropriate.

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