Runs CI tests locally to rapidly identify and fix code failures.

Install

mkdir -p .claude/skills/ci-loop && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/14737" && unzip -o skill.zip -d .claude/skills/ci-loop && rm skill.zip

Installs to .claude/skills/ci-loop

Activation

This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.

Run the full CI pipeline locally and iterate until all tests pass.
66 chars · catalog descriptionno explicit “when” trigger
Intermediate

Key capabilities

  • Run server tests using Gradle.
  • Run shared module tests (JVM/server) using Gradle.
  • Run Android tests using Gradle.
  • Parse test failures to extract class, method, error, and stack trace.
  • Apply fixes iteratively for failing tests.
  • Re-run specific tests after applying fixes.

How it works

This skill automates the local CI loop by running various test suites (server, shared, Android), parsing any failures to identify the source, applying fixes iteratively, and re-running specific tests until all tests pass. It concludes with a final verification step.

Inputs & outputs

You give it
Code changes or a codebase with potential test failures
You get back
All tests passing, no compilation errors

When to use ci-loop

  • Run CI locally
  • Debug failed tests
  • Verify fix iteration

About this skill

CI Loop Skill

Run the full CI pipeline locally and iterate until all tests pass.

Trigger

Use this skill when:

  • Starting work on a feature or bugfix
  • Before creating a commit or PR
  • When asked to "run tests" or "make sure everything passes"
  • After making changes to verify they work

Workflow

1. Initial Assessment

# Check current state
./gradlew assemble --dry-run

2. Run Full Test Suite

# Server tests
./gradlew :server:test --info

# Shared module tests (JVM/server)
./gradlew :shared:serverTest --info

# Android tests
./gradlew :clients:android:app:testDebugUnitTest --info

3. Parse Failures

For each failure in the output:

  1. Extract the test class and method name
  2. Extract the error message and stack trace
  3. Identify the source file and line number

4. Fix Loop

while failures exist:
    for each failure:
        1. Read the failing test file
        2. Read the source file under test
        3. Analyze the failure using failure-patterns.json
        4. Apply the fix
        5. Re-run the specific test:
           ./gradlew :module:test --tests "FullyQualifiedTestClass.testMethod"
        6. If still failing, try alternative fix
        7. If fixed, move to next failure

5. Final Verification

# Run full suite to catch regressions
./gradlew check

# If all pass, report success
# If new failures, add to queue and continue

Quick Commands

# Run specific test
./gradlew :server:test --tests "dev.hyo.deepline.server.routes.UserRoutesTest.testCreateUser"

# Run tests with stacktrace
./gradlew :server:test --stacktrace

# Continue after failure
./gradlew :server:test --continue

Success Criteria

  • All tests pass (exit code 0)
  • No compilation errors
  • Security validation passes (see security-scan skill)

When not to use it

  • When the project does not use Gradle.
  • When not working on a feature or bugfix.
  • When all tests already pass.

Limitations

  • The skill is designed for Gradle-based projects.
  • It relies on specific Gradle commands for different test suites.
  • Security validation is mentioned but not detailed within this skill's scope.

How it compares

This skill provides an automated, iterative workflow for debugging and fixing failing tests locally, simplify the process of achieving a passing CI pipeline compared to manual debugging.

Compared to similar skills

ci-loop side by side with the closest alternatives in the catalog.

SkillInstallsUpdatedSafetyDifficulty
ci-loop (this skill)04moReviewIntermediate
springboot-verification44moReviewIntermediate
code-checklist04moReviewBeginner
semoss-testing-ci02moNo flagsIntermediate

Try saying

Example prompts that trigger this skill in your AI assistant.

Search skills

Search the agent skills registry