Jerry Lin
Community@ledtorch · Taiwan / San Francisco / Tokyo
Status: Overloading… | A Taiwan based product designer who works for a fintech company and love to share the case study from practical cases.
Agent Skills by Jerry Lin
Showing 10 vetted skills indexed across 1 GitHub repositories.
ledtorchcommit
Proposes grouped git commits with conventional messages and waits for user confirmation.
ledtorchinspect
Audits named code components for duplication, dead code, and logic bugs, producing severity-labelled findings and a plan document.
ledtorchhandoff
Generates a verified session handoff document with git fingerprint, shipped work, and next tasks.
ledtorchplan
Decomposes specs into ordered, verifiable implementation tasks with acceptance criteria.
ledtorchsetup
Prepares a project's references documentation layer for agent workflow commands.
ledtorchlabel
Renames Figma frame layers to role-based PascalCase design-system names.
ledtorchrelease
Cuts a git release end-to-end: clean tree, push, tag, release notes, deploy.
ledtorchtrim
Trim a named file to its smallest form without losing actionable facts or changing behavior.
ledtorchteam
Orchestrates parallel sub-agent teams through a gated plan-run-review pipeline.
ledtorchaeo
Audits websites for AI-crawler access, raw-HTML visibility, and citation readiness, producing severity-labelled findings and a plan.
Frequently Asked Questions About Jerry Lin
FAQPage SchemaWhat tasks can I perform with ledtorch's skills?▼
You can create convention-following git commits, audit components for duplication and dead code, break specs into ordered tasks, implement plan-doc tasks with patch snapshots, trim files to minimal form, rename Figma layers, cut tagged releases, generate session handoff docs, and audit sites for AI-crawler readability.
Who are these skills designed for?▼
Software engineers and product designers who run structured, command-driven development sessions. The git, release, inspect, plan, and build skills target developers, while the label skill serves designers maintaining design-system naming conventions in Figma, and aeo serves site owners checking answer-engine visibility.
How do I set up and run these skills in a project?▼
Run install.sh, then invoke the setup skill to either wire commands into an existing doc layer (adopt) or build that layer from the real codebase (init). Afterward, commands like /commit, /plan, /build, and /release operate against the project, reporting if references/ is missing.
Do the skills modify code automatically without approval?▼
No. The inspect and aeo skills only produce severity-labelled findings and plan docs, fixing nothing. The build skill implements tasks but never commits, leaving changes in the working tree, and the team skill enforces two review gates with a maximum of two rollbacks.
What does the aeo skill check on a website?▼
It audits Answer/Agent Engine Optimization: AI-crawler access, raw-HTML visibility, answer-first content structure, citation-worthiness, freshness, and entity signals. It outputs severity-labelled, evidence-tiered findings plus a plan doc, indicating whether AI assistants can read, chunk, and cite the site.