skillZs
★ LIVE SKILL TAGS ★
>>> LIVE SKILLS INDEX <<<
* OPEN SOURCE *
NO LOGIN, NO TRACKING
※ REAL INSTALL DATA ※
← back to all skills
ohyeh/agent-scripts232 installs

using-skills

Which skill owns this task. Invoke BEFORE doing by hand any of: .pdf read or fill, test plan, architecture improvement, second-model review, docs writing, greenfield page, prose cleanup, 「有沒有 skill」, 「用哪個」 — or any task not named in the kernel routing index. Even a 1% chance a skill covers it means invoke this first; it costs one read.

How do I install this agent skill?

npx skills add https://github.com/ohyeh/agent-scripts --skill using-skills
view source ↗

Is this agent skill safe to install?

  • Gen Agent Trust Hubpass

    The skill is a central routing directory designed to help an AI agent delegate tasks to specialized skills based on user intent. It includes logic for determining task ownership and provides a maintenance command for verifying configuration consistency. No malicious behaviors, obfuscation, or security risks were identified.

  • Socketpass

    No alerts

  • Snykwarn

    Risk: MEDIUM · 1 issue

What does this agent skill do?

using-skills

Route by intent. Every name below is an entry in skills-lock.json, the installed roster, so a name here always resolves on this machine.

The kernel routing index in ~/.claude/CLAUDE.md binds first. When the kernel names a trigger — delegation, done/stuck claims, unclear acceptance, retries, loop-shaped work, code changes, output shape — its route wins. A few of those skills still get a row below so that every lock name resolves in one place; the row is a lookup, not a second opinion.

Two hops maximum: using-skills → domain router → member. Never route back here from inside a domain router. Read the live SKILL.md before acting; never act from memory of an old roster.

Mode column: Skill() = model-invocable · manual = read its SKILL.md inline or use its slash command · via-router = enter through the router named.

Hop 1: domain routers

RouterEnter it whenMode
using-design-skillsAnything visual: web page, app screen, HTML deliverable, chart, artifact, motion, plus module and API interface designSkill()
using-workflowsLoop-shaped work: audit, consensus, triage, plan→build, lifecycle, research rounds. Owns the recipes in ~/.claude/workflows/; this file does not list themSkill()
using-tmux-agent-toolsRunning a CLI as a tmux worker, or deciding inline versus workerSkill()

Flows: situation → handoff → artifact → loop

Conditional handoffs, not a pipeline. Use only the stages the task needs, inside the scope the user authorized. A domain router that owns a pipeline keeps it. Loop-shaped work or a named recipe goes to using-workflows, which picks or bypasses; a fresh reviewer alone does not make a recipe. Not every skill is a link: tools, generators and constraint sets come from the map.

SituationHandoffArtifact per stepLoop closes when
New feature, acceptance unclearunknowns-discovery (when its trigger holds) → prototype for a design question, or wayfinder when too big for one session → implement (tdd when the user wants test-first) → a check executed and quoted before any done claim (kernel Done rule) → defect-first-review when risk or the user asksstated defaults → design answer or decision tickets → diff and tests → quoted check output → findingsfindings fixed and re-verified; a repeated finding → a gate, or a backlog row that names the file
Bug or regressiondiagnosing-bugs (many competing hypotheses → enter using-workflows instead) → re-run the original scenario → defect-first-review on the fix diff when risk warrantsrepro and failing signal → fix and regression test → original scenario re-run → findingsregression test stays in the repo
Test planwrite the cases inline → only when execution is authorized: agent-browser (web) or agent-device (app) → a bug → Bug row; tracker issues → triagecases → execution evidence → bug reporta missed case goes back into the plan
Pause or hand offsession-handoff; reusable knowledge for Codex → shared-memory-intake (inbox only)handoff doc → inbox entrythe next session starts from the doc

Owner pointers (their chains stay with them): codebase-wide deepening → improve-codebase-architecture; interface, domain or visual design → using-design-skills; writing for people → writing-artifacts; tmux workers → using-tmux-agent-tools; loops and recipes → using-workflows; a skill edit → skill-creator (draft, eval), then rule maintenance §1 (exact diff, approval) before any install or deploy.

Shape an idea before building

I need to…SkillMode
diverge in parallel under different cognitive framesadhdSkill()
stress-test a plan or decision in frontier roundsgrillingSkill()
pick a flow when the situation is fuzzy or cross-domainask-novamanual
surface the map/territory gap in unfamiliar territoryunknowns-discoverySkill()
build a throwaway prototype to answer a design questionprototypeSkill()
chart work too big for one session as decision ticketswayfindermanual

Write and change code

I need to…SkillMode
find deepening opportunities across a whole codebaseimprove-codebase-architecturemanual
design a deep module interfacecodebase-designSkill()
build or sharpen the domain model, CONTEXT.md, an ADRdomain-modelingSkill()
build features test-firsttddSkill()

Diagnose and verify

I need to…SkillMode
diagnose a hard bug or performance regression inlinediagnosing-bugsSkill()
review a diff defect-first, read-only, every findingdefect-first-reviewSkill()
review UI code against Web Interface Guidelinesweb-design-guidelinesSkill()

Delegate and orchestrate

I need to…SkillMode
write any worker brief (GOAL/ACCEPTANCE/REPORT)delegation-templatesSkill()
drive tmux workers: mechanics and wrapperstmux-agent-toolsvia-router
plan and run an explicitly orchestrated agent workflowcodex-dynamic-workflowsSkill()

Read the outside world

I need to…SkillMode
read a pasted URL as clean markdowndefuddleSkill()
read, fill, merge, or produce a PDFpdfSkill()
investigate a question against primary sources, write it upresearchSkill()
drive a browser: navigate, fill forms, screenshotagent-browserSkill()
drive an iOS, Android, macOS, or TV appagent-deviceSkill()
read or drive the Grok Bot macOS app (bots, folders, transcripts)using-grok-bot-appSkill()

Write for people to read

I need to…SkillMode
turn a fuzzy subject into a finished deliverable (entry point)writing-artifactsSkill()
write in-repo software docs: README, API, tutorialdocumentation-writingSkill()
mine raw fragments before any structurewriting-fragmentsmanual
shape raw material into an article paragraph by paragraphwriting-shapemanual
assemble material into a journey of beatswriting-beatsmanual
remove AI writing patterns from any prosestop-slopSkill()

Visual and HTML output (all through using-design-skills)

Listed so a name resolves, not as a bypass. The router picks the owner.

SkillOwns
impeccableDirection authority: product UI, landing and marketing pages, redesigns, polish, critique
apple-designSprings, gestures, interruptible motion
hallmarkAnti-slop greenfield pages, audits, design extraction from a URL or screenshot
diagram-designHouse-style typed diagrams, mermaid and draw.io, PNG/SVG export
data-reportCSV, Excel, or JSON into a visual report page

Session and fleet upkeep

I need to…SkillMode
hand off to a fresh sessionsession-handoffSkill()
curate shared Codex memory, or submit findings to itshared-memory-intakeSkill()
create, edit, or eval a skillskill-creatorSkill()
move issues and external PRs through triage rolestriagemanual
generate a bash wizard for steps only a human can dowizardSkill()

Plugin and bundled skills the fleet relies on

Not in skills-lock.json. Listed only when the plugin is enabled in ~/.claude/settings.json and the skill shows in the available-skills listing; a disabled plugin's skills do not belong here. Mode plugin:<name> = lives in that plugin's cache; bundled = ships with Claude Code.

SkillSourceMode
context-modecontext-modeplugin:context-mode
ctx-doctorcontext-modeplugin:context-mode
ctx-indexcontext-modeplugin:context-mode
ctx-insightcontext-modeplugin:context-mode
ctx-purgecontext-modeplugin:context-mode
ctx-searchcontext-modeplugin:context-mode
ctx-statscontext-modeplugin:context-mode
ctx-upgradecontext-modeplugin:context-mode
ponytailponytailplugin:ponytail
ponytail-auditponytailplugin:ponytail
ponytail-debtponytailplugin:ponytail
ponytail-gainponytailplugin:ponytail
ponytail-helpponytailplugin:ponytail
ponytail-reviewponytailplugin:ponytail
artifact-designClaude Codebundled
artifact-capabilitiesClaude Codebundled
artifact-diagrammingClaude Codebundled
workflow-authoringClaude Codebundled
loopClaude Codebundled
claude-apiClaude Codebundled

Other enabled plugins (code-review, commit-commands, pr-review-toolkit, session-report, frontend-design, security-guidance) show in the listing with their own descriptions. Rules files (~/.agents/rules/*.md) are not skills; the kernel routes them.

Subagent exemption

A worker executing one assigned task does not enter this router. The dispatcher already routed; the worker follows its brief.

Red flags

Naming one and proceeding anyway needs a stated reason.

  • 「這只是小問題,不用 skill」: go direct when the domain is obvious, but do not skip routing on substantial work.
  • 「我記得那支 skill 的內容」: members change. Read the live SKILL.md first.
  • 「先做完再回頭套流程」: for loop-shaped or visual work the router picks the process first. Retrofitting ships half-done work.

Freshness self-check

This map matches skills-lock.json at its last edit. When a name here does not resolve, or the listing shows a skill this map omits, confirm before trusting it:

diff <(rg -o '`[a-z][a-z0-9-]+`' ~/.agents/skills/using-skills/SKILL.md | tr -d '`' | sort -u) \
     <(jq -r '.skills|keys[]' ~/.agents/.skill-lock.json | sort)

Names in the left column only are the plugin/bundled section, mode words, rule names, or stale references; check the ones outside that section. Names in the right column only are installed skills this map has not placed yet.

Nothing fits

Not loop-shaped, not visual, no intent above: check the active available-skills listing, then work inline. A new recurring intent goes to the user as a proposal for a skill or recipe. Never improvise a router inline.

Add the canonical catalog link to the repository README so users can inspect current installs and available audits. The publishing guide covers the complete discovery path.

<a href="https://skillzs.dev/skills/ohyeh/agent-scripts/using-skills">View using-skills on skillZs</a>