Critical Thinking

Evaluator last updated September 3, 2026.

Overview

The Critical Thinking evaluator rates a student’s source-based argumentative essay for critical thinking, treated as distinct from writing quality. Given the essay, the assignment prompt, and the source passage(s) the student was given, it returns a rating for 5 indicators:

Evidence strength
Synthesizing sources (if more than one source is provided)
Counterarguments
Facts over opinions
Drawing conclusions

The evaluator also returns an “Overall” rating, with evidence and reasoning included for all ratings.

At a glance


Inputs	Student essay, assignment prompt, source passage(s), and source count
Supported grades	8–10
Rubric	Project rubric adapted from the Carnegie Foundation / ETS Skills Progressions, informed by the SFF critical-thinking framework

The evaluator was built and validated using the model below (other configurations, including cheaper models, produced lower and less stable agreement — see Validation and limitations):


Model used	Claude Opus 5, version-pinned (`claude-opus-5-20260723`)
Temperature	Not configurable — Opus 5 accepts no sampling parameters

Getting started

Follow the Quickstart to start using this evaluator:

Access method
Evaluators playground	View in Platform ↗
SDK	Not yet available (v0.1, early access)
GitHub	Python notebook ↗ Prompts (`system.txt` ↗ and `user.txt` ↗) Schemas (`input_schema.json` ↗ and `output_schema.json` ↗)

Inputs

Input	Description	Required
Assignment text	Writing prompt the student answered	Yes
Source passage(s)	Source passage(s) shown to the student	Yes
Source count	Number of source passages provided (rates “Synthesizing sources” only when there is more than 1 passage)	Yes
Essay text	Student essay (200–800 words recommended)	Yes

Output

Field	Description
Indicators	One entry per rated indicator — Synthesizing sources, Evidence strength, Counterarguments, Facts over opinions, Drawing conclusions
Indicator evidence	Verbatim quote(s) from the essay per indicator, each with a short comment on what it demonstrates
Indicator reasoning	Reasoning for the rating, written before the rating itself
Indicator rating	5-level scale: Not Evident, Exploring, Analyzing, Integrating, Extending
Overall reasoning	Reasoning including the median arithmetic across the indicator ratings
Overall rating	Headline Critical Thinking rating, same five-level scale

Interpreting results

Output	How to use
Indicator rating + evidence/reasoning	Formative signal on where an essay sits on a specific critical-thinking skill Example: Flag essays that rate low on Counterarguments or Facts over opinions for targeted instruction; show the quotes so a teacher can check the rating
Overall rating	Cohort- or portfolio-level signal on critical thinking, distinct from writing quality Example: Research or program-level analysis of critical-thinking trends across a set of essays, rather than an individual grade

Evaluator release history

Date	Changed
September 3, 2026	First release

Durable skills evaluators

Explore evaluators that rate student work against a construct’s indicators.

Quickstart

Run an evaluator in the Evaluators playground, a Python notebook, or with the SDK.

Understanding evaluators

Getting started

SDK API reference

Student-Facing Text evaluators

Feedback evaluators

Academic standards evaluators

Durable skills evaluators

Datasets

Resources

Overview

At a glance

Getting started

Inputs

Output

Interpreting results

Evaluator release history

Durable skills evaluators

Quickstart

​Overview

​At a glance

​Getting started

​Inputs

​Output

​Interpreting results

​Evaluator release history

​Related topics

Durable skills evaluators

Quickstart

Overview

At a glance

Getting started

Inputs

Output

Interpreting results

Evaluator release history

Related topics