Geeigneter Einsatz: Ideal for AI agents that need wrap development branch. Lokalisierte Zusammenfassung: A little project to make an internet controlled lamp! This AI agent skill supports Claude Code, Cursor, and Windsurf workflows.

How do I install wrap?

Run the command: npx killer-skills add mt-krainski/twinkletaps/wrap. It works with Cursor, Windsurf, VS Code, Claude Code, and 19+ other IDEs.

Which IDEs are compatible with wrap?

This skill is compatible with Cursor, Windsurf, VS Code, Trae, Claude Code, OpenClaw, Aider, Codex, OpenCode, Goose, Cline, Roo Code, Kiro, Augment Code, Continue, GitHub Copilot, Sourcegraph Cody, and Amazon Q Developer. Use the Killer-Skills CLI for universal one-command installation.

Are there any limitations for wrap?

Einschraenkung: After fixes, stage and commit locally (git add + agent-utils git-commit). Do not push yet — Step 6 handles the push.. Einschraenkung: Do not proceed to Step 6 until every test is green.. Einschraenkung: Every claim in Step 7's PR description must be backed by command output..

wrap

A little project to make an internet controlled lamp! This AI agent skill supports Claude Code, Cursor, and Windsurf workflows. Wrap Development Branch

SKILL.md

Readonly

Upstream Repository Material

The section below is imported from the upstream repository and should be treated as secondary evidence. Use the Killer-Skills review above as the primary layer for fit, risk, and installation decisions.

Supporting Evidence

Wrap Development Branch

Name: wrap
Availability: InStock
Author: mt-krainski

The single exit point for all code changes. Every change that gets pushed goes through /wrap.

Automated pipeline: review -> fix -> lint -> test -> commit -> PR -> Jira -> CI watch.

Announce at start: "Wrapping up — running review -> fix -> lint -> test -> commit -> PR -> Jira -> CI watch."

Resuming after interruption: If you were previously interrupted during /wrap (e.g. session crash, context overflow), start the entire pipeline from Step 1. Do not attempt to resume from where you left off — re-running all steps is safer and ensures nothing is skipped.

Pipeline

Step 1: Determine Context

bash
1git branch --show-current

bash
1git merge-base HEAD origin/<base-branch>

bash
1git rev-parse HEAD

Determine the base branch from branch history (usually main).

Infer Jira issue key from branch name (e.g. task/GFD-42/slug -> GFD-42). If found, read the issue with jira-utils get-issue for context.

Step 2: Code Review

Follow the /review skill to spawn the code-reviewer agent. Provide context from Step 1 (SHAs, issue description, acceptance criteria).

Wait for the review result.

Step 3: Apply Review Fixes

Severity	Action
Critical	Fix immediately. If unclear, ask.
Important	Fix before proceeding.
Minor	Fix if quick (<2 min). Otherwise note and skip.

If no Critical or Important issues, skip to Step 4.

After fixes, stage and commit locally (git add + agent-utils git-commit). Do not push yet — Step 6 handles the push.

Step 4: Lint

bash
1npm run lint

Fix lint errors introduced by our changes. Pre-existing unrelated issues can be ignored.

Step 5: Test

No completion claims without fresh evidence.

bash
1npm run test
2npm run test:e2e

ALL tests must pass. Your starting assumption is that all tests were passing before your changes. Any test failure is caused by your changes until you have exhaustively proven otherwise (e.g. by checking out the base branch and reproducing the failure there).

Investigate every failure. Read error messages, screenshots, and test code.
Fix whatever your changes broke — including tests in other files that depend on behavior you changed (e.g. ARIA roles, API contracts, CSS classes).
Never dismiss a failure as "unrelated" without concrete proof.
Never remove or weaken a test to make it pass.
Do not proceed to Step 6 until every test is green.

Every claim in Step 7's PR description must be backed by command output.

Step 5b: UI Screenshot Review (Advisory)

Only runs when UI changes are detected.

Check if the diff contains .tsx files under src/components/ or src/app/:
```
bash
1git diff --name-only <merge-base>..HEAD -- '*.tsx'
```
Filter for paths matching src/components/ or src/app/. If none found, skip this step.
Glob for screenshots in test-results/screenshots/: If no screenshots exist, skip this step.
Spawn a visual-qa subagent to review the screenshots:
- Provide paths to all screenshots found
- The subagent checks for structural breakage: layout overflow, collapsed containers, clipped content, blank screens
- This is advisory only — findings do NOT block the pipeline
Report findings to the user. Do not fail the pipeline based on visual review results.

Step 6: Commit & Push

Invoke the /commit skill. Squash any fixup commits before the final commit.

This stages, commits, and pushes — but does NOT open a PR (that's Step 7).

Step 7: Open PR

bash
1agent-utils gh-pr-create --base <target-branch> --title '[GFD-###] <Title>' --body '<body>'

Target branch: Usually main. Ask if unclear.
PR body: Task ID, "What changed" (brief bullets), "How to test", "Depends on" if relevant.
If a PR already exists for this branch: skip PR creation. Just push.

Return the PR URL.

Step 8: Update Jira

If a Jira issue was found in Step 1:

Transition to Review: jira-utils transition-issue --issue-key <KEY> --transition-id 2
Read humanAtlassianId from .workflow, assign to human via jira-utils update-issue --issue-key <KEY> --assignee "<humanAtlassianId>"
Add comment via jira-utils add-comment: summary of implementation + PR link + "Please review and merge, or leave feedback."

Do NOT transition to Done. Done only after PR is merged to mainline.

Step 9: CI Watch

Only runs if a PR was created/updated and changes were pushed.

Use the Agent tool to spawn a background sub-agent (run_in_background: true) with this prompt:

Poll CI for PR #<PR_NUMBER> every 60 seconds using agent-utils gh-pr-checks <PR_NUMBER>. If all checks pass, report success. If any check fails, investigate with agent-utils gh-run-view <RUN_ID> and agent-utils gh-run-view --log-failed <RUN_ID> and report the failure details. Do not poll more than 30 times (30 minutes). If still pending after that, report timeout.

Do NOT use sleep to poll. The background agent handles the waiting. You will be notified when it completes.

While CI runs, report to the user: "CI running in background — you'll be notified when it completes."

When the background agent returns:

All checks pass: Report success and finish.
Any check fails: Investigate, fix the issue, commit, push, and spawn a new CI watch agent.

Failure Modes

Review finds critical architectural issues: Stop. Report to user. Tests fail and can't be fixed after thorough investigation: Stop. Report. Don't create PR with failing tests. No Jira issue found: Skip Jira steps. Derive PR description from git log. CI fails and can't be fixed after 3 attempts: Stop. Report the failure details to the user.

Red Flags

Never:

Skip the review step
Create PR with failing tests
Claim tests pass without running them
Force-push without explicit request

Always:

Run lint and tests fresh before PR
Include evidence of passing tests
Stop and ask if review reveals fundamental issues

wrap — for Claude Code twinkletaps, community, for Claude Code, ide skills, bash git branch --show-current, bash git rev-parse HEAD, GFD-42, jira-utils get-issue, review, git add

Über diesen Skill

Funktionen

# Core Topics

Killer-Skills Review

Warum diese Fähigkeit verwenden

Am besten geeignet für

↓ Handlungsfähige Anwendungsfälle for wrap

! Sicherheit & Einschränkungen

Why this page is reference-only

Source Boundary

Decide The Next Action Before You Keep Reading Repository Material

Start With Installation And Validation

Cross-Check Against Trusted Picks

Move To Workflow Collections For Team Rollout

Browser Sandbox Environment

⚡️ Ready to unleash?

FAQ & Installation Steps

? Frequently Asked Questions

What is wrap?

How do I install wrap?

What are the use cases for wrap?

Which IDEs are compatible with wrap?

Are there any limitations for wrap?

↓ How To Install

! Reference-Only Mode