rogerchappel

Validate, report on, pack, and locally install reusable agent skill folders.

Review code changes for correctness, safety, maintainability, and test coverage.

Generate focused tests for deterministic local software changes.

too short

Review code quality and suggest improvements for pull requests

Help debug errors by analyzing stack traces and suggesting fixes

Scan code for security vulnerabilities and hardening opportunities

A well-formed example skill for testing skillpack validation

Operate, explain, troubleshoot, or automate the LaunchClip CLI for repository, product, topic, voiceover, presenter, downloaded HeyGen-avatar, and promotion-packet video workflows. Use when an agent needs to choose LaunchClip commands, prepare inputs, run or resume the model-directed production pipeline, inspect costs and artifacts, validate a workspace, render after approval, or use the legacy dry-run packet lane. Distinguish API-backed CLI production from the separate subscription-agent workflow.

Create an end-to-end HyperFrames video with the current subscription agent as the creative and orchestration layer. Use when a user wants a video, promo, explainer, product story, repository walkthrough, or social clip from a URL, repository, topic, brief, script, supplied media, or a downloaded HeyGen avatar while avoiding LaunchClip's metered model pipeline and not depending on HyperFrames plugin skills. Ask a compact intake, author the editable composition, verify it locally, obtain preview approval, and render the final file.

Manage tasks, projects, agents, and inbox via the CrewCmd API. Use when creating, updating, listing, or dispatching tasks; managing projects; checking agent status; sending inbox items; or any task/project board operations. Replaces any external task board (Mission Control, Jira, Linear, etc.) when CrewCmd is the configured workspace backend. Triggers on: task creation, task updates, project management, agent dispatch, inbox operations, board queries, "create a task", "what's queued", "update the task", "mark as done".

Convert messy brain dumps, PRDs, voice transcripts, GitHub issues, and planning notes into deterministic agent-ready task briefs or queues. Use this when the user asks to split rough work into repo-scoped tasks, generate docs/TASKS.md from a PRD, prepare work for Codex/OpenClaw/Claude Code/GitHub Copilot, or create Markdown/JSON/YAML task queues.

Audit local agent tool manifests and explain capability risk before enabling tools.

Generate local-first security checklist action plans from reviewed JSON profiles.

Use when helping a user research roles, assess fit, prepare applications, or plan follow-up during a job search without inventing credentials or submitting personal data.

Author, lint, test, render, and package portable coding-agent skills.

Research Meta/Facebook ads, develop high-performing ad concepts, generate asset prompts, prepare safe draft ads, and report performance.

Use cutpilot to turn local footage into simple transcript-aware EDLs, 15-second shorts, and carousel-style social edits with ffmpeg dry-run planning and pre-render checks.

Create, refine, preview, and print greeting cards with the Paperhug CLI on an OpenClaw host. Use when a user asks to make a card, revise a card draft, preview a printable card, or send a card to a configured local printer.