.agents/skills/fix-ci/SKILL.md
Run CI checks, triage lint/type/test failures, apply fixes in bounded batches, and verify until `bun run ci-check` is green.
npx skillsauth add TheCardGoat/lorcana-engine fix-ciInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Use this skill when CI is failing at the monorepo level and you need to make focused fixes quickly while keeping changes scoped and verified.
the-card-goat-online repository.bun run ci-check (runs format, lint, type-check, tests, build).node_modules.From the repository root:
bun run ci-check
If this passes, no changes are required.
Classify output into:
format: formatter/reporting issueslint: pattern path/to/file.ts:line:col lint/rule-name ...type: pattern path/to/file.ts(line,col): error TSxxxx: ...test: pattern FAIL path/to/test.ts or failing assertion tracesGroup by file first, then by package.
If the same file repeatedly appears and no safe fix is obvious, document it as blocked and continue with the next item.
For each targeted file/package:
Package-wide validation (preferred):
bunx turbo lint check-types test --filter=@tcg/<package>
If package checks are noisy, run narrower checks in sequence:
bunx turbo lint --filter=@tcg/<package>
bunx turbo check-types --filter=@tcg/<package>
bunx turbo test --filter=@tcg/<package>
After each batch:
bun run ci-check after local convergence.If an issue remains after 3 correction attempts for the same scope, mark it as blocked and move on.
Report:
file or package: error-type - what was changedfile or package: reason + suggested workaroundbun run ci-checkany just to satisfy a type check.documentation
One-time setup that gathers design context for your project and saves it to your AI config file. Run once to establish persistent design guidelines.
tools
Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Claude's capabilities with specialized knowledge, workflows, or tool integrations.
development
--- name: review-pr-comments description: Systematically process GitHub pull request review feedback end-to-end: pick a target PR, identify unanswered or unresolved comments, handle each thread sequentially, implement required code updates, run repository verification checks, and post clear responses. Use when asked to review PR comments, resolve review threads, or address reviewer suggestions/questions. --- # Review PR Comments Process pull request feedback one thread at a time and finish wit
testing
Tone down overly bold or visually aggressive designs. Reduces intensity while maintaining design quality and impact.