Where to Find the Response Quality Rubric for i‑have‑adhd

The response quality rubric for the i‑have‑adhd repository is located at evals/rubric.md.

This markdown file defines the evaluation dimensions and scoring criteria used to assess skill output before any release. If you're contributing to or auditing the i‑have‑adhd skill, understanding this rubric is essential for ensuring your changes meet the project's quality standards.

What the Response Quality Rubric Contains

The evals/rubric.md file enumerates five core scoring dimensions with defined weights:

• Correctness – factual accuracy of responses
• Autonomy – appropriate delegation vs. independent action
• Actionability – concrete, usable guidance provided
• Safety – avoidance of harmful or risky recommendations
• Concision – clarity without unnecessary verbosity

The rubric also specifies release conditions that must be satisfied before a candidate version can be promoted.

How to Access the Rubric

View Directly in the Repository

cat evals/rubric.md

Load Programmatically for Custom Evaluation Scripts

import pathlib

rubric_path = pathlib.Path(__file__).parent.parent / "evals" / "rubric.md"
rubric_text = rubric_path.read_text(encoding="utf-8")
print(rubric_text)

This pattern is useful when building automated evaluation pipelines that reference the canonical rubric.

How the Rubric Is Referenced

The response quality rubric is not an isolated document. According to the ayghri/i-have-adhd source code, it is cross-referenced in:

• evals/README.md – documents the evaluation harness and points to rubric.md as the authoritative scoring standard
• skills/i-have-adhd/SKILL.md – the canonical skill definition whose output is measured against the rubric's criteria

This ensures consistency: any modification to the skill flows through evaluation protocols that rely on the same rubric.

Related Files

File	Purpose
evals/rubric.md	Definitive response quality rubric
evals/README.md	Evaluation harness documentation
skills/i-have-adhd/SKILL.md	Skill definition under evaluation

Summary

• The response quality rubric for i‑have‑adhd lives at evals/rubric.md.
• It governs five dimensions: Correctness, Autonomy, Actionability, Safety, and Concision.
• Access it via cat, raw git read, or programmatic loading in Python.
• Evaluation workflows in evals/README.md and skill definitions both defer to this single source of truth.

Frequently Asked Questions

What format is the response quality rubric in?

The rubric is a markdown file (.md) designed for human readability and programmatic parsing. The five evaluation dimensions are described with their relative weights and specific criteria for passing scores.

Can I modify the rubric for my own evaluation?

Yes. Since the rubric is version-controlled in the repository, you can fork ayghri/i-have-adhd and adapt evals/rubric.md to your needs. Be sure to update any references in evals/README.md if you change the file's structure or location.

How is the rubric used in practice?

The evaluation harness documented in evals/README.md loads and applies rubric.md to score candidate versions of the i‑have‑adhd skill. Outputs from skills/i-have-adhd/SKILL.md are assessed against the rubric's dimensions, and releases are blocked until all release conditions are satisfied.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}