- Home
- Skills
- Testing & Debugging
- Smart Test Runner: Run Only the Tests You Affected
More screenshots
Works with the AI Agents you already use
Smart Test Runner: Run Only the Tests You Affected
Turn a slow test suite into just the tests your change can break: maps your git diff to the affected tests and prints the exact command to run.
$19
About this skill
Your suite takes minutes, so after every edit the agent either skips the tests or burns the session waiting for all of them. On networkx, one changed file (algorithms/cycles.py) needs 5 of 268 test files, and Test Impact Selector names them in seconds, each with the import chain that justifies it.
It ships a real tool, impact.py (standard-library Python, nothing to install), that reads your repository's dependency graph instead of guessing from file names:
- Python: imports (relative too), names traced through
__init__.pyre-exports andimport *,conftest.pyfan-in,mock.patch("pkg.mod.f")strings, doctests. - JS/TS: import/export/require/dynamic import,
vi.mock/jest.mock, tsconfigpaths, barrel files, workspace packages, vitest/jest config globs. - Go, Java/Kotlin, C#: packages, imports, namespaces and type names.
- The exact command for your runner:
pytest,vitest run,jest --runTestsByPath,go test ./pkg,mvn -Dtest=...,dotnet test --filter. Nearest tests come first. - A safe fallback: when a root config, lockfile, unknown file or unsupported language changes, the verdict is FULL (exit 1) and it prints the full-suite command.
- NOT COVERED: changed code that no test reaches is listed, so nobody claims it was tested.
- validate_selection.py: seeds bugs and reverts real commits on your own repo, runs your full suite for ground truth, and reports whether the selection held every newly failing test.
Measured (pinned commits, numbers in the screenshots):
| Repo | Change | Selected |
|---|---|---|
| networkx (976 files) | cycles.py | 5 of 268 test files |
| networkx, seeded bug | flow/capacityscaling.py | 20 of 268, every newly failing test inside |
| networkx, seeded bug | temporalisomorphvf2.py | 268 of 268 (its conftest really reaches everything), all inside |
| date-fns (2,070 files) | addDays/index.ts | 45 of 324 test files |
| demo fixture | seeded rounding bug | 5 of 8 selected; the 3 selected pytest files catch it |
2 of 2 seeded breaking changes contained, 0 misses: a small sample, so the tool to measure your own repo is included.
What it does not do: it is static analysis, so it cannot see reflection, dependency-injection containers or plugins loaded by computed names. Tightly coupled code gives big selections (django utils/text.py: 619 of 642). It never replaces the final full run: the verdict always asks for one before "done".
Works in Claude Code, Cursor, Codex CLI, Copilot and any SKILL.md agent.
Privacy: runs locally and read-only; no network, no telemetry. Only --run and validate_selection.py --apply execute your tests.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Testing & Debugging.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 27 days ago
- Software
Pythonvitest
Node.jsjest
Git