A personal collection of custom AI Agent Skills compatible with gh skill install. These skills extend AI coding assistants with specialized tools, instructions, and workflows.
| Skill Name | Description | Installation Command | Compatible Agents |
|---|---|---|---|
commit-it |
Commit changes following the Conventional Commits specification, verifying test passage first. | gh skill install handlename/agent-skills commit-it |
Claude Code, Copilot, Gemini CLI, Cursor |
create-pr |
Create a Pull Request from the current branch to a base branch, using GitHub CLI to generate a suitable title and description from the commit history. For structural changes, attaches a big-picture diagram as an SVG image (drawn via mermaid-to-svg and uploaded via github-attachment-upload, mermaid source kept in a collapsed details block, falling back to an inline mermaid block). |
gh skill install handlename/agent-skills create-pr |
Claude Code, Copilot, Gemini CLI, Cursor |
draw-mermaid |
Draw a mermaid diagram from a natural-language intent and return a syntax-validated mermaid source, verified with mermaid.parse() against a vendored, version-pinned mermaid.js run in a browser via playwright-cli (offline, deterministic). Designed to be called from other skills (e.g. plan-to-issue, create-pr) through a subagent. |
gh skill install handlename/agent-skills draw-mermaid |
Claude Code, Copilot, Gemini CLI, Cursor |
github-attachment-upload |
Upload an arbitrary local file to GitHub user-attachments by driving a browser with playwright-cli against an issue editor, and return the attachment URL (no issue is created). File-type agnostic (SVG, PNG, screenshots, …). Designed to be called from other skills (e.g. plan-to-issue, create-pr) through a subagent. Runs @playwright/cli via npx (no preinstall needed); requires a one-time GitHub sign-in in its persistent browser profile. |
gh skill install handlename/agent-skills github-attachment-upload |
Claude Code, Copilot, Gemini CLI, Cursor |
github-resource-access |
Rules for accessing GitHub resources with the gh CLI: reads are free, writes require an explicit instruction. Includes a quick reference of read vs. write commands, red flags for unrequested changes, and exceptions for skills whose purpose is a write operation. |
gh skill install handlename/agent-skills github-resource-access |
Claude Code, Copilot, Gemini CLI, Cursor |
go1.27-changes |
Reference for what changed in Go 1.27 — read when writing or reviewing Go code in a go 1.27+ module so pre-1.27 training-data assumptions don't creep in. Covers generic methods, new/changed standard-library APIs (encoding/json/v2, uuid, crypto/mldsa, strings.CutLast, …), and dangerous behavior changes (unbuffered time channels, UnixConn raw io.EOF, localhost-only go tool trace -http), each linked to official docs with a verify-before-use rule. |
gh skill install handlename/agent-skills go1.27-changes |
Claude Code, Copilot, Gemini CLI, Cursor |
hwwwn |
Passive reference for what information belongs where and how to phrase code comments. HWWWN routes each kind of information to its channel — How to code, What to test code, Why to the commit log, Why not to code comments (after t_wada) — then gives concrete comment rules: keep it minimal, break lines only at sentence boundaries (no artificial column limit), and never restate what the code shows (a long explanatory comment signals a responsibility that should be separated instead). Language/project conventions take precedence. | gh skill install handlename/agent-skills hwwwn |
Claude Code, Copilot, Gemini CLI, Cursor |
maint-glossary |
Create and maintain a well-structured glossary of domain terms from a requirements document, spec, or codebase, through collaborative refinement — extracting, categorizing, and defining terms for a fresh glossary, and diffing the source to add/revise/deprecate entries when maintaining an existing one. | gh skill install handlename/agent-skills maint-glossary |
Claude Code, Copilot, Gemini CLI, Cursor |
mermaid-to-svg |
Turn a mermaid diagram into a hand-drawn free-layout SVG image file and return its path. Pure and offline (no mermaid-cli, no browser, no upload). Designed to be called from other skills (e.g. plan-to-issue, create-pr) through a subagent, then paired with github-attachment-upload to publish the SVG. |
gh skill install handlename/agent-skills mermaid-to-svg |
Claude Code, Copilot, Gemini CLI, Cursor |
multirepo |
Investigate or edit multiple already-cloned local git repositories in parallel. Fires on a multirepo: <repo1>, <repo2>, ... declaration; the main session orchestrates one multirepo-worker sub-agent per repo (dispatch and result aggregation only), creating worktrees for edits and leaving commits/PRs to other skills. |
gh skill install handlename/agent-skills multirepo |
Claude Code |
plan-to-issue |
Turn a finalized implementation plan or design into a GitHub tracking issue, with a big-picture diagram posted as an SVG image (drawn via mermaid-to-svg and uploaded via github-attachment-upload, mermaid source kept in a collapsed details block, falling back to an inline mermaid block), a PR-granularity checklist (TASK- IDs), plus a self-contained comment that lets a context-less person or agent start implementing. |
gh skill install handlename/agent-skills plan-to-issue |
Claude Code, Copilot, Gemini CLI, Cursor |
setup-tagpr |
Set up tagpr for automated release management: creates and updates release pull requests for unreleased items, tags them on merge, and creates GitHub Releases. Covers goreleaser (Go) and TypeScript/Node.js projects. | gh skill install handlename/agent-skills setup-tagpr |
Claude Code, Copilot, Gemini CLI, Cursor |
This repository is also a Claude Code plugin marketplace.
Add the marketplace once, then install the bundled plugin:
claude plugin marketplace add handlename/agent-skills
claude plugin install handlename@handlenameAll skills listed under Available Skills are bundled into a single handlename plugin, together with the worktree-setup SessionStart hook. The repository root itself is the plugin (.claude-plugin/marketplace.json points at source: "."), so the plugin reads the same skills/<name>/ directories that gh skill install fetches. Both paths share one physical source with no symlinks and no duplicated files.
| Component | Type | Description |
|---|---|---|
commit-it |
skill | Commit changes following the Conventional Commits specification, verifying test passage first. Supports both git and Jujutsu (jj) repositories. |
create-pr |
skill | Create a Pull Request from the current branch to a base branch, using GitHub CLI to generate a suitable title and description from the commit history. |
draw-mermaid |
skill | Draw a mermaid diagram from a natural-language intent and return a syntax-validated mermaid source, checked with mermaid.parse() against a vendored, version-pinned mermaid.js via playwright-cli. Designed to be called from other skills through a subagent. |
github-attachment-upload |
skill | Upload an arbitrary local file to GitHub user-attachments via playwright-cli (through an issue editor, no issue created), returning the attachment URL. File-type agnostic. Designed to be called from other skills through a subagent. |
github-resource-access |
skill | Rules for accessing GitHub resources with the gh CLI: reads are free, writes require an explicit instruction. |
go1.27-changes |
skill | Reference for Go 1.27's developer-facing changes (language, standard library, go command) with gotchas foregrounded, for writing or reviewing Go code without falling back on pre-1.27 training-data assumptions. |
hwwwn |
skill | Passive reference for what information belongs where and how to phrase code comments: HWWWN routes How to code, What to test code, Why to the commit log, and Why not to code comments (after t_wada), then gives concrete comment rules — minimal, line breaks only at sentence boundaries, and no comment that restates what the code shows. Language/project conventions take precedence. |
maint-glossary |
skill | Create and maintain a glossary of domain terms from a requirements document, spec, or codebase — extracting, categorizing, and defining terms for a fresh glossary, and diffing the source to add/revise/deprecate entries when maintaining an existing one. |
mermaid-to-svg |
skill | Turn a mermaid diagram into a hand-drawn free-layout SVG image file and return its path. Pure and offline (no mermaid-cli, no browser, no upload). Designed to be called from other skills through a subagent, then paired with github-attachment-upload. |
multirepo |
skill | Investigate or edit multiple already-cloned local git repositories in parallel via one multirepo-worker sub-agent per repo. The main session only orchestrates; worktrees are created for edits, and commits/PRs are left to other skills. |
persona-mimic-trainer |
skill | Train and iteratively refine a custom sub-agent that mimics a specific person's decision-making criteria, cognitive biases, and communication style for mock wall-hitting (pre-communication verification). |
plan-to-issue |
skill | Turn a finalized implementation plan or design into a GitHub tracking issue (overview + PR-granularity checklist plus a self-contained detailed comment), or post it as a comment on an existing related issue. |
setup-tagpr |
skill | Set up tagpr for automated release management: creates and updates release pull requests for unreleased items, tags them on merge, and creates GitHub Releases. Supports goreleaser (Go) and TypeScript/Node.js projects. |
worktree-setup |
hook | At session start (SessionStart hook), copies untracked files declared in .worktreeinclude from the main git worktree into linked worktrees, and runs setup scripts declared in .worktreesetup (e.g. mise trust, aqua policy allow) for repositories on a trust allowlist. |
To install a skill from this repository, run the following command using the GitHub CLI:
gh skill install handlename/agent-skills <skill-name>gh skill install handlename/agent-skills commit-itWhen you install a skill using gh skill install, the GitHub CLI:
- Clones/fetches the specific skill directory under
skills/<skill-name>/from this repository. - Injects tracking and version metadata directly into the frontmatter of your local
SKILL.md. - Places the skill into the appropriate directory depending on your scope:
- Project Scope (Default): Installed inside your current repository at
.agents/skills/<skill-name>/ - User Scope: Installed in your home directory (e.g.,
~/.gemini/skills/or similar, depending on your agent configuration).
- Project Scope (Default): Installed inside your current repository at