Agent-Skills.md

Agent Skills: Testing instructions

When creating tests and using tests for testing of application operation. Use this skill any time you are writing, running, or verifying tests — including pytest unit tests, integration tests, terminal verification scripts, or browser-based E2E flows. Trigger even for partial requests like "run the tests", "write a test for X", "verify this works", "check that the sync works", or "does this pass?". If any form of testing or verification is involved, use this skill. Load with read_file on .kilocode/skills/testing/SKILL.md (ignore the absolute path in the location tag).

UncategorizedID: ScotterMonk/AgentAutoFlow/testing

Repository

ScotterMonk/AgentAutoFlow
ScotterMonkLicense: CC0-1.0
371

Install this agent skill to your local

pnpm dlx add-skill https://github.com/ScotterMonk/AgentAutoFlow/tree/HEAD/.kilocode/skills/testing

Skill Files

Browse the full folder contents for testing.

Download Skill

Loading file tree…

.kilocode/skills/testing/SKILL.md

Skill Metadata

Name
testing
Description
When creating tests and using tests for testing of application operation. Use this skill any time you are writing, running, or verifying tests — including pytest unit tests, integration tests, terminal verification scripts, or browser-based E2E flows. Trigger even for partial requests like "run the tests", "write a test for X", "verify this works", "check that the sync works", or "does this pass?". If any form of testing or verification is involved, use this skill. Load with read_file on .kilocode/skills/testing/SKILL.md (ignore the absolute path in the location tag).

Testing instructions

For creating, running, and verifying tests. Read skill-local guidance only when .kilocode/skills/testing/AGENTS.md is confirmed to exist; otherwise use root AGENTS.md for project-specific commands, paths, and environment notes.

Constraint: Do not execute multi-line Python scripts directly in the terminal.

Hard rule — silent stdout means write a script:

  • If python -c "..." produces no terminal output, do NOT keep tweaking the same one-liner. After 2 silent runs (first = warning, second = trigger), STOP and save it as a .py file in an appropriate project folder, then run that file.
  • Multi-line python -c invocations are forbidden inline. Even when they appear to work, output can be unreliable in some terminal environments.
  • Pipelines that mix python -c with Select-String, Out-File, etc. can also drop stdout. Same rule: write a script.
  • Quick shell sanity checks (python --version, path existence checks, single short prints) are fine and do not count.

Procedure for Multi-line Scripts:

  1. Search: Check codebase and memory for existing scripts.
  2. Evaluate:
    • Exact Match Found: Execute the existing script.
    • Similar Match Found:
      • Analyze dependencies (what relies on this script?).
      • Determine strategy: Modify existing vs. Duplicate new.
      • Execute the modified or duplicated script.
    • No Match: Create a new script file, then execute it.

Testing types

Index: Terminal Scripts, Python Tests, Browser.

Terminal Scripts

One-line tests: Run directly in the terminal. Multi-line tests:

  • Search the codebase and memory to see if an equivalent script already exists.
    • If exact: reuse it.
    • If similar: create or modify an appropriate .py script.
  • Run the script from its .py file instead of pasting multiple lines.

Python tests

Framework: Use the project's existing Python test framework. Test locations: Use the project's established test folders.

Running tests: Use the commands documented in local AGENTS.md or the project root guidance.

Creating tests:

  • Use the existing test library and project fixtures.
  • Do not add database setup/teardown unless the project has a database and the test requires it.
  • Tests should be self-contained; prefer temporary paths and fixtures over real project paths.
  • Create tests so that success/green yields only the test name and "pass" for output — no descriptions or explanations in the output.

Browser

  • Use browser-use skill for E2E flows.
  • If logging in to this application's site, use login-using-querystring skill.