Claude Code Plugins

Community-maintained marketplace

Feedback

validation-workflow

@get2knowio/maverick
0
0

Use this skill to validate code changes pass all checks (format, lint, build, tests). Runs validation script and iteratively fixes failures. Invoke before finalizing PRs or after making code changes.

Install Skill

1Download skill
2Enable skills in Claude

Open claude.ai/settings/capabilities and find the "Skills" section

3Upload to Claude

Click "Upload skill" and select the downloaded ZIP file

Note: Please verify skill by going through its instructions before using it.

SKILL.md

name validation-workflow
description Use this skill to validate code changes pass all checks (format, lint, build, tests). Runs validation script and iteratively fixes failures. Invoke before finalizing PRs or after making code changes.

Validation Workflow

This skill provides a structured process for validating code changes and fixing any failures.

When to Use

  • After implementing features or fixes
  • After addressing code review feedback
  • When /fly or /refuel reaches the validation phase
  • When explicitly asked to verify the build passes

Phase 1: Run Validation

Send notification: Run ${CLAUDE_PLUGIN_ROOT}/scripts/notify.sh testing "Running validation"

Execute the validation script:

${CLAUDE_PLUGIN_ROOT}/scripts/run-validation.sh

Parse the JSON results to determine status of each check:

  • Format check
  • Lint check
  • Build
  • Tests

Phase 2: Handle Results

If All Checks Pass

Report success:

Validation Status:
- Format check: ✅ pass
- Lint check: ✅ pass
- Build: ✅ pass
- Tests: ✅ pass (X passed)

Proceed to next workflow phase.

If Any Check Failed

  1. Send notification: Run ${CLAUDE_PLUGIN_ROOT}/scripts/notify.sh error "Validation failed"

  2. Parse error output from failed checks

  3. Create TODO list of failures

  4. Fix failures in priority order:

    • Compilation errors (highest priority)
    • Linting errors
    • Test failures
    • Formatting issues (lowest priority)
  5. Fix ALL failures - even if unrelated to current changes

Phase 3: Fix Failures

For Compilation/Lint Errors

Address each error directly based on the error message. Common fixes:

  • Missing imports
  • Type mismatches
  • Unused variables
  • Clippy warnings

For Test Failures

Spawn subagents for each failing test:

Fix failing test: [test name]
Error: [error output]
File: [location]

Investigate whether it's a test bug or implementation bug.
Fix the actual issue - do NOT weaken assertions.

For Formatting Issues

cargo fmt --all

Phase 4: Iterate

After fixing issues:

  1. Re-run validation:

    ${CLAUDE_PLUGIN_ROOT}/scripts/run-validation.sh
    
  2. Check results

  3. If still failing: Return to Phase 3

  4. Maximum iterations: 5 attempts before reporting blockers

Phase 5: Commit Fixes

If any changes were made to fix validation issues:

git add -A
git commit -m "fix: resolve validation failures

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>"
git push

Output

After completing validation, report:

## Validation Status

- Format check: ✅/❌ pass/fail
- Lint check: ✅/❌ pass/fail
- Build: ✅/❌ pass/fail
- Tests: ✅/❌ pass/fail (X passed, Y failed)

### Issues Fixed
- [List of issues that were fixed during validation]

### Remaining Blockers (if any)
- [List of issues that could not be resolved after 5 iterations]