AgentSkillsCN

verify-and-fix

运行格式化、静态代码分析工具、单元测试和构建检查，然后修复违规问题。

SKILL.md
--- frontmatter
name: verify-and-fix
description: Run formatting, linters, unit tests, and build checks, then fix violations.

Verify and Fix

Purpose

Automate the process of ensuring code quality, functional correctness, and buildability within the Terraform Provider Lightdash project. This skill provides a comprehensive local verification loop using trunk, make, and pre-commit.

Safety & Restrictions

  • •NO Acceptance Tests: Do NOT run make testacc or any command that sets TF_ACC=1. This skill is intended for fast, local feedback loops (linting, building, unit testing) and should not interact with live infrastructure.

Workflow

1. Formatting

Start by ensuring consistent code formatting across the repository.

  • •Command: trunk fmt --all
  • •Action: Run this command to automatically format all files. If any changes are made, they will be reflected in the working directory.

2. Linting

Execute the project's linting suite to identify violations.

  • •Command: make lint
  • •Details: This command runs trunk check --all and pre-commit run --all-files.
  • •Analysis: Review the output for any errors or warnings.

3. Unit Testing

Execute the project's unit tests to ensure functional correctness.

  • •Command: make test
  • •Details: This runs go test on the internal packages without setting TF_ACC.
  • •Action: Verify that all tests pass.

4. Automated Fixing

If linter violations are detected, attempt to resolve them automatically.

  • •Condition: Only if make lint reports fixable issues.
  • •Command: trunk check --all --fix
  • •Action: This will apply automated fixes provided by the linters integrated into Trunk.

5. Build Verification

Ensure the project compiles successfully and passes security/deadcode checks.

  • •Command: make build
  • •Details: This triggers gen-docs, go-tidy, gosec, deadcode, and finally go build.
  • •Action: Verify that the build completes without errors.

6. Iterative Manual Fixes

If issues remain after automated attempts, proceed with manual intervention.

  • •Analyze: Examine the specific error messages from trunk check, make test, or make build.
  • •Fix: Use the Edit tool to address the root causes of the violations in the source code.
  • •Verify: Re-run the loop starting from Step 2 (Linting) until all checks pass and the build is successful.

Termination Criteria

  • •make lint returns no errors.
  • •make test returns no failures.
  • •make build completes successfully.
  • •No remaining linter violations are reported by trunk check --all.

Examples

Scenario: Fixing a Bug Found by Unit Tests

  1. •Format: trunk fmt --all (Success).
  2. •Lint: make lint (Success).
  3. •Unit Testing: make test (Fails in internal/lightdash/api/utils_test.go).
  4. •Manual Fix:
    • •Analyze the test failure.
    • •Edit internal/lightdash/api/utils.go to fix the logic.
  5. •Verify: Re-run make test. (Success).
  6. •Build: make build. (Success).