# Testing And Evaluation

> Instructions for changes verification. It's a mandatory read when 1. asked for running any kind of tests, or 2. Finished implementing a milestone - always after finishing a plan.

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

---


# Required Verification steps

- Cheap steps, do after finishing a modification task:
  - typecheck of modified packages
  - prettier of modified files
- After every bigger step:
  - lint of modified packages
  - Unit tests of modified packages
- After reaching a milestone or finishing plan implementation:
  - if GUI part was changed, run playwright integration tests from app/gui
  - always, regardless of what was changed: electron package tests from
    app/electron-client, including AI tests.

# How to run Electron Package tests

Run `git submodule update --init --recursive` before the build. Without it
`--mode staging` silently falls back to defaults and the resulting binary fails
the very first e2e step (login screen never renders, console shows
`Failed to fetch`).

The ide build should be with staging environment by default (`--mode staging`
option in case of `./run ide build` script). The credentials file should be
provided by user. If working on a worktree, try to copy it from main repo.