Pytest Testing Framework

Python testing

by HoangNguyen0403 · MIT license · ★ 569 on GitHub

Test Python services with pytest, async coverage, monkeypatch, and boundary-focused fakes. Use when writing Python tests, fixtures, async tests, regression tests, or dependency-isolated verification.

---
name: python-testing
description: Test Python services with pytest, async coverage, monkeypatch, and boundary-focused fakes. Use when writing Python tests, fixtures, async tests, regression tests, or dependency-isolated verification.
metadata:
  triggers:
    files:
      - "tests/**/*.py"
      - "pytest.ini"
      - "**/conftest.py"
      - "**/test_*.py"
      - "**/*_test.py"
    keywords:
      - pytest
      - monkeypatch
      - fixture
      - async test
      - regression test
---

Overview

Test Python services with pytest, async coverage, monkeypatch, and boundary-focused fakes. Use when writing Python tests, fixtures, async tests, regression tests, or dependency-isolated verification.

Manual install (npx, ZIP)

Claude Codenpx skills add HoangNguyen0403/agent-skills-standard --skill python-testing -g -a claude-code

Claude Code (degit)npx degit HoangNguyen0403/agent-skills-standard/skills/python/python-testing#develop ~/.claude/skills/python-testing

Claude.ai / desktopZip the folder and upload it under Settings → Capabilities → Skills.

Codex, Cursor, other agentsPoint them at SKILL.md. The other files are plain text.

Connects with Agent Alley MCP · agentalley.io/mcp

Files included

SKILL.md3.3 KB

Read SKILL.md here

HoangNguyen0403/develop1 file shown

SKILL.md3.3 KB

Browse every file on GitHub↗

SKILL.md

Show the full text63 lines

python-testing/SKILL.md63 lines · 3.3 KB

PreviewCode

Outline

RawView on GitHub

Python Testing

Priority: P0 (CRITICAL)

Core Rule Anchors

[BE-TEST-01] Parametrized Tests for Equivalent Inputs: Prefer pytest.mark.parametrize for equivalent inputs and boundary permutations. Distinct scenarios may stay separate.
[BE-TEST-02] Ban on Brittle SQL String Matching: Test pure domain calculations directly or use fakes/testcontainers; do not regex-match SQL strings in mocks.
[BE-TEST-03] Ban on Shallow Assertions: Never assert only assert result without inspecting domain attributes and invariants.
[BE-TEST-04] Ban on Pass-Through Seam Mocks: Repositories and adapters must test contract compliance and error mapping; ban 1:1 pass-through mock echoing.
[BE-TEST-05] Bug-First Regression Lock: Every PR fixing a bug ticket or with title fix(...) must introduce a failing test reproducing the defect before fixing it.

Rules

Use pytest and pytest-asyncio for async boundaries.
Add regression tests around contract parsing, workflow routing, and verifier outcomes ([BE-TEST-05]).
Prefer pytest.mark.parametrize for equivalent inputs; distinct scenarios may stay separate ([BE-TEST-01]).
Prefer fakes or monkeypatch at external seams: DB, RPC, Telegram, subprocess, Docker ([BE-TEST-02], [BE-TEST-04]).
Verify behavior and domain invariants, not shallow truthiness or implementation trivia ([BE-TEST-03]).
Coverage is diagnostic and project-configured; verify risk-weighted critical paths rather than padding code for an arbitrary percentage.

Recipe

Write the failing test first for the boundary or regression ([BE-TEST-05]).
Patch one seam high enough to keep the test readable ([BE-TEST-04]).
Assert structured outcomes: metadata, route result, blocker text, emitted command ([BE-TEST-03]).
Use async tests for async code; avoid spinning real event loops manually.
Cover negative paths for malformed packets, stale runtime state, and missing env.

Anti-Patterns

[BE-TEST-01] Stylistic parametrize rewrites: Do not demand parameterized rewrites of passing tests without a behavioral gap.
[BE-TEST-02] SQL regex matching: Do not regex-match SQL strings in mocks.
[BE-TEST-03] Shallow assertions: Never assert only assert result; always verify domain attributes and invariants.
[BE-TEST-04] Pass-through seam mocks: Ban 1:1 mock echoing without contract assertions.
[BE-TEST-05] Bug fix without reproduction test: Ban bug fixes without a reproduction test.
No live network or DB in unit tests: Use fakes or testcontainers.
No broad monkeypatch spray: Patch the boundary, not every helper under it.
No smoke-only proof for shared runtime changes: Add focused assertions too.

References

1	---
2	name: python-testing
3	description: Test Python services with pytest, async coverage, monkeypatch, and boundary-focused fakes. Use when writing Python tests, fixtures, async tests, regression tests, or dependency-isolated verification.
4	metadata:
5	triggers:
6	files:
7	- "tests/*/.py"
8	- "pytest.ini"
9	- "**/conftest.py"
10	- "*/test_.py"
11	- "*/_test.py"
12	keywords:
13	- pytest
14	- monkeypatch
15	- fixture
16	- async test
17	- regression test
18	---
19
20	# Python Testing
21
22	## Priority: P0 (CRITICAL)
23
24	## Core Rule Anchors
25
26	- `[BE-TEST-01]` Parametrized Tests for Equivalent Inputs: Prefer `pytest.mark.parametrize` for equivalent inputs and boundary permutations. Distinct scenarios may stay separate.
27	- `[BE-TEST-02]` Ban on Brittle SQL String Matching: Test pure domain calculations directly or use fakes/testcontainers; do not regex-match SQL strings in mocks.
28	- `[BE-TEST-03]` Ban on Shallow Assertions: Never assert only `assert result` without inspecting domain attributes and invariants.
29	- `[BE-TEST-04]` Ban on Pass-Through Seam Mocks: Repositories and adapters must test contract compliance and error mapping; ban 1:1 pass-through mock echoing.
30	- `[BE-TEST-05]` Bug-First Regression Lock: Every PR fixing a bug ticket or with title `fix(...)` must introduce a failing test reproducing the defect before fixing it.
31
32	## Rules
33
34	- Use `pytest` and `pytest-asyncio` for async boundaries.
35	- Add regression tests around contract parsing, workflow routing, and verifier outcomes (`[BE-TEST-05]`).
36	- Prefer `pytest.mark.parametrize` for equivalent inputs; distinct scenarios may stay separate (`[BE-TEST-01]`).
37	- Prefer fakes or monkeypatch at external seams: DB, RPC, Telegram, subprocess, Docker (`[BE-TEST-02]`, `[BE-TEST-04]`).
38	- Verify behavior and domain invariants, not shallow truthiness or implementation trivia (`[BE-TEST-03]`).
39	- Coverage is diagnostic and project-configured; verify risk-weighted critical paths rather than padding code for an arbitrary percentage.
40
41	## Recipe
42
43	1. Write the failing test first for the boundary or regression (`[BE-TEST-05]`).
44	2. Patch one seam high enough to keep the test readable (`[BE-TEST-04]`).
45	3. Assert structured outcomes: metadata, route result, blocker text, emitted command (`[BE-TEST-03]`).
46	4. Use async tests for async code; avoid spinning real event loops manually.
47	5. Cover negative paths for malformed packets, stale runtime state, and missing env.
48
49	## Anti-Patterns
50
51	- `[BE-TEST-01]` Stylistic parametrize rewrites: Do not demand parameterized rewrites of passing tests without a behavioral gap.
52	- `[BE-TEST-02]` SQL regex matching: Do not regex-match SQL strings in mocks.
53	- `[BE-TEST-03]` Shallow assertions: Never assert only `assert result`; always verify domain attributes and invariants.
54	- `[BE-TEST-04]` Pass-through seam mocks: Ban 1:1 mock echoing without contract assertions.
55	- `[BE-TEST-05]` Bug fix without reproduction test: Ban bug fixes without a reproduction test.
56	- No live network or DB in unit tests: Use fakes or testcontainers.
57	- No broad monkeypatch spray: Patch the boundary, not every helper under it.
58	- No smoke-only proof for shared runtime changes: Add focused assertions too.
59	## References
60
61	- [Framework Map](../references/framework-map.md)
62	- [Pytest Patterns](references/pytest-patterns.md)
63