# Evaluate Plugin

> How to verify the Cognitive OS is functioning correctly.

- Skill: `attac-t/evaluate-plugin` (Agent Skill)
- Install (CLI): `npx skillmds@latest add attac-t/evaluate-plugin`
- Raw SKILL.md: https://api.skillmd.com/api/skills/attac-t/evaluate-plugin/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: attac-t (https://skillmd.com/u/attac-t)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/attac-t/evaluate-plugin

---


# Skill: Evaluate Plugin

> "Trust, but verify."

## 1. The Principles
*   **Behavioral Testing**: Do not just check for syntax. Check for *mindset*.
*   **Isolation**: Test the Brain (Memory), Muscles (Agents), and Reflexes (Hooks) separately.

## 2. The Test Suite
### A. The Reflex Test (Hooks)
*   **Action**: Create a file that violates a convention (e.g., a file in the wrong location, or with forbidden content).
*   **Expectation**: The OS should immediately flag the violation (if configured in `hooks.json`).

### B. The Brain Test (Memory)
*   **Action**: Run `/blueprint`.
*   **Expectation**: It should load the blueprint.

### C. The Muscle Test (Agents)
*   **Action**: Run `/design "A simple cache capability"`.
*   **Expectation**: The Architect should *Plan* before *Doing*. It should use the `design` style.

### D. The Identity Test (Version)
*   **Action**: Ask the host which version of this plugin it loaded. Read the version in the plugin's own manifest.
*   **Expectation**: They match. **Report both numbers, never a verdict alone** — a session that answers "current" without saying against what has proved nothing.
*   **Why**: A, B and C all pass on an old copy. They test the plugin that is running, and never whether it is the plugin that was shipped.

## 3. Execution
Use the `/evaluate` command to run this suite interactively.