القائمة
Automated test-driven fix loop — run tests, analyze failures, investigate root causes, fix code, and re-run until all tests pass with configurable max iterations
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
Create or update lightweight phased development plan documents with checkbox tasks, test cases, and completion criteria for implementation work
Dual-mode review skill — review implementation plans before coding (plan mode) or review code changes after implementation (code mode) with architecture, risk, and test analysis
Plan and coordinate parallel development workstreams using Claude Code sub-agents with worktree isolation for safe concurrent implementation
Generate or update release documentation — README.md styling with badges and navigation, CHANGELOG entries, and release notes for version management
استنادا إلى تصنيف SOC المهني
| name | test-fix-loop |
| description | Automated test-driven fix loop — run tests, analyze failures, investigate root causes, fix code, and re-run until all tests pass with configurable max iterations |
| allowed-tools | ["Read","Edit","Write","Glob","Grep","Agent","Bash(npm test:*)","Bash(npx:*)","Bash(bun test:*)","Bash(bun run:*)","Bash(pytest:*)","Bash(python3 -m pytest:*)","Bash(go test:*)","Bash(cargo test:*)","Bash(jest:*)","Bash(vitest:*)","Bash(make test:*)","Bash(make:*)","Bash(tsc:*)","Bash(git diff:*)","Bash(git status:*)","Bash(ls:*)","Bash(wc:*)"] |
| arguments | ["target"] |
| argument-hint | [test command, file pattern, or 'investigate' for analysis-only mode] |
| user-invocable | true |
| when_to_use | Use when the user wants to run tests and fix failures in a loop, debug test failures, investigate root causes, or achieve passing tests through iterative fixes. Examples: 'run tests and fix', 'test fix loop', 'make tests pass', 'fix failing tests', 'investigate failure', '테스트 돌리고 고쳐줘', '테스트 통과시켜', '실패 원인 분석', '테스트 루프', '디버그' |
테스트를 실행하고, 실패를 분석하고, 코드를 수정하고, 다시 실행하는 루프를 자동화한다. 원인 분석(investigate) 모드도 포함한다.
$target: 다음 중 하나
npm test, pytest tests/api/)tests/auth/*.test.ts)investigate — 수정 없이 원인 분석만$target이 명령어가 아니면 자동 감지한다. 상세 감지 규칙과 실패 분류는 references/tfl-workflow-guide.md를 참조한다.
| 파일 | 프레임워크 | 기본 명령어 |
|---|---|---|
package.json (scripts.test) | npm/bun | npm test / bun test |
vitest.config.* | Vitest | npx vitest run |
jest.config.* | Jest | npx jest |
pyproject.toml [tool.pytest] | pytest | python3 -m pytest |
go.mod | Go test | go test ./... |
Cargo.toml | Cargo | cargo test |
Makefile (test target) | Make | make test |
Bash로 테스트를 실행하고 전체 출력을 캡처한다.
출력에서 실패 정보를 추출한다:
| 분류 | 패턴 |
|---|---|
| Compilation error | SyntaxError, TypeError, cannot find module, build 실패 |
| Assertion failure | AssertionError, expect(...).toBe(...), assertEqual 실패 |
| Runtime exception | ReferenceError, NullPointerException, segfault |
| Timeout | timeout, exceeded, SIGTERM |
실패한 테스트 이름, 파일 위치, 에러 메시지, 스택 트레이스를 구조화한다.
Investigation 모드 ($target = investigate): 이 단계에서 멈추고 분석 결과를 보고한다.