agent-skills
claudeCodex code review, multi-agent debate, multi-agent deep research, overbuild check, build/verify effort loop
6 skills
How to decide, review, write to people, run releases and work with a team
Plugin workstyle from marketplace mikhin in mikhin/claude-plugins. Commands read from the repo's manifest — 9 days ago.
This plugin runs code on your machine — its hooks fire automatically on events like tool use. The exact commands are listed below; read them before installing.
all-checkRun every check the project has except e2e, scoped to the files of the current PR or branch, and report one line per check. Triggers — "/all-check", "run all the checks", "all checks except e2e", "check the PR files".
answerA teammate or the client asked the user something — find the question, say what is actually asked, check earlier decisions and the code, contract and tickets, and draft the reply. Never sends. Triggers — "/answer", "<name> asked me", "reply to <name>", "what do I answer", "I don't get what they ask"
buyPick one thing to buy — criteria, the user's marketplaces in order, real reviews, original or copy — and give the top three with one pick. Never orders or pays. Triggers — "/buy", "what should I buy", "find me a replacement", "original or copy", "search the marketplace".
commsHow to write to people outside the code on the user's behalf — clients, product owners, teammates, product users. Draft first, never send; no blame, no hinted defects, no promises; one question instead of a pile of comments; copy-paste ready without quote marks, a Slack draft goes straight to the cl
contract-impactShow how an API contract change and the frontend affect each other — what a contract PR or spec change forces in the frontend, or what a UI or design needs that the contract lacks. Triggers — "/contract-impact", "does the frontend need changes", "is the contract ready for this UI", "does the spec ma
decisionsHow to make decisions with the user — the questions to answer before proposing a change, where to look answers up, and what to reject on sight. Meant to be always on (SessionStart hook), not trigger-loaded.
deskPut the result of the session — txt, png, pdf, screenshots or a draft — in the user's output folder, in the subfolder of its topic, and answer with the path. Triggers — "/desk", "save it to the desktop", "put a txt on the desktop", "screenshots to the desktop", "save the result".
figma-diffCompare a PR's UI with its Figma frames, state by state and at every viewport, and put paired screenshots plus a findings list on the Desktop. Triggers — "/figma-diff", "compare with the design", "does it match Figma", "compare the PR with Figma", "how will it look in the release".
nextName the user's one next step on the current project — what blocks others first (reviews, unanswered mentions, stuck release issues), then the user's own work. Triggers — "/next", "what's next", "next step", "what should I do now".
org-researchResearch one topic across the whole project — every repo of the GitHub org (code on every branch, issues, PRs, review threads), all of Linear (issues, comments, documents, projects) and the project's Slack channels and DMs — and write a dated, linked txt report to the Desktop. Read-only. Triggers —
pendingTurn the user's thought about a PR into comments in their pending (unsubmitted) GitHub review — English, friendly, anchored to lines — and never submit unless the user says publish. Triggers — "/pending", "add to pending", "leave a pending comment", "put it in the pending review", "publish the pendi
pinPin this session so `pins` in the terminal resumes it from any directory
pruneTidy git in the project repos — prune, drop merged branches and stray worktrees, fast-forward to the active release branch. Never touches unmerged work. Triggers — "/prune", "clean up git", "tidy the repos".
pulseOne teammate's activity day by day across GitHub, Linear and Slack, compared with their own usual pace and their release scope — evidence, not judgement — and a non-toxic way to ask. Triggers — "/pulse <name>", "what did <name> do", "is <name> slow", "<name>'s activity", "how do I ask without being
release-closeCheck a release by facts, not statuses — every issue of the Linear project against its PRs in every project repo, everyone's issues included — and list the tails with an action for each. Read-only until the user acks. Triggers — "/release-close", "is the release clean", "what is left in the release"
reportDraft the user's project update for the client channel covering every working day since the last one, from GitHub, Slack and Linear activity. Draft only, never posts. Triggers — "/report", "write my update", "daily report", "report for the last days".
reviewReview the current branch, a PR or a commit in Claude Code or Codex — runs an independent review at medium, then judges every finding against the code (accepted / rejected / ticket / comment) and ends with a merge verdict. Also how bugs get investigated — the "why didn't the tests catch it" rule, te
slack-copyPut a drafted Slack message on the macOS clipboard so it pastes with working links and code formatting. Never sends. Triggers — "/slack-copy", "copy for slack", "pbcopy this for slack", "with links for slack", "put it in the clipboard for slack".
stackHow work gets verified — the frontend checking stack expected on every change, how layout is reviewed across viewports, and what matters on a product the user runs. Meant to be always on (SessionStart hook), not trigger-loaded.
stage-testTest a release on staging with two separate subagents — one verifies release readiness across Slack, Linear, GitHub and deployments and saves a TXT test plan; the other executes it and reports release blockers. Triggers — "/stage-test", "test it on staging", "is it safe to merge to prod", "we merged
talkToggle voice mode — replies are spoken aloud by macOS `say` and kept to one or two short conversational sentences. Needs the voice hooks from the README. Triggers — "/talk", "/talk on", "/talk off", "voice mode", "turn the voice on", "turn the voice off".
trade-offA hard client date and work that does not fit — find what the client needs on that date, the cheapest path to it (hardcode, manual step, cut scope) with its debt written down, and draft the proposal to the team. Client business before code purity. Triggers — "/trade-off", "we won't make it", "what c
"${CLAUDE_PLUGIN_ROOT}/hooks/autoload.sh"
Codex code review, multi-agent debate, multi-agent deep research, overbuild check, build/verify effort loop
6 skills
Warns when session context grows past a threshold and offers a handoff
hooks
Sends grep for code symbols to the LSP tool
hooks
Paces a weekly model quota at 1/7 per day
hooks
Sends a long answer back once to be rewritten as a tldr
hooks
Snap a frame from your iPhone camera above the prompt and send it to Claude with your next message
Talk like caveman. Cut filler. Keep technical details explicit.
22 skills · 1 command · 3 agents · hooks
Persistent memory for Claude Code: the next session starts where the last one left off.
29 skills · 1 MCP server · hooks
tasteskill 2.5.1: professional company websites at the figma.com and x.ai level, with one section per screen, real photographs, real product UI, effects kept for tech and gaming pages, and a real-browser checker and finish gate.
1 skill
Open-source AI job search that runs locally in your AI coding CLI: scan job portals, evaluate listings into a structured A-H report with a 1-5 score, tailor your CV, track applications
5 skills · hooks
Cross-session memory and token savings for coding agents.
6 skills · 1 agent · 1 MCP server · hooks
Package any website or local web build into a lightweight desktop app with Pake CLI.
1 skill
This page updates itself from your repo. Add the badge to your README, or get it Featured to pin it above the organic results with a followed link to your site.