Comments
Code comments and annotations
revkit
by konradmichalik
Use revkit for GitHub/GitLab PR review operations — reading a PR, listing, filtering, replying to and resolving review comments, checking CI status, and fetching failure logs. Prefer it over raw gh/glab API calls: it returns compact, normalized JSON that costs far fewer tokens than parsing raw API responses, and it works identically on GitHub and GitLab.
review-github-comments
by hoopit
Handle all review comments on a GitHub PR. Use when the user asks to fix or handle a PR or PR comments.
refactor-archeology
by ShotaIuchi
Code history and rationale investigation for refactoring. Apply when researching git blame, commit history, code comments, hidden constraints, and historical design decisions.
upgrade-setup
by medTech-group
Move a project from solo mode to the full set-up, one upgrade at a time. "Protect main" makes a ruleset and the merge settings on the base branch with gh after one confirmation on the owner's own computer, or guides the person click by click to them everywhere else, reads them back with workflow protection, and opens the pull request that leaves solo mode. "Require these checks", once CI has run on the base branch, adds the ticked checks that ran to the ruleset protecting it (with "up to date" on) and to the profile's ci.required_checks. "Set up unattended runs" asks the questions only runs need, guides the two cloud environments, and writes the answers to the profile in one pull request. "Add try-it links" takes the owner through Cloudflare one screen at a time, then opens a pull request that turns previews on in the profile and reports its own try-it link once it has built. Use when the person asks to protect main, to require the checks, to set up unattended runs or to add try-it links, or picks one where the kit offers it.
code-review
by magfrump
Orchestrate a comprehensive code review by coordinating code-fact-check and code critic agents (security-reviewer, performance-reviewer, api-consistency-reviewer) in parallel, with optional contextual critics (architecture-review, test-strategy, tech-debt-triage, dependency-upgrade, ui-visual-review) auto-selected based on the diff. Follows a 3-stage pipeline: code fact-check → critic agents → synthesis. Produces a freeform chat summary plus a structured code review rubric with red/amber/green status tracking. Use this skill when the user asks to "review this code", "review this PR", "review my changes", "review this diff", "full code review", "run all critics", "code review this branch", or wants a multi-perspective review of code changes. Default to this orchestrator whenever a PR is being prepared, opened, or evaluated — it composes security, performance, API consistency, and (when triggered) architecture into a single pass. Also use when the user asks for two or more of those concerns together. For a deliberately narrow review on a single concern (just security, just performance), invoke the standalone critic skill directly instead.
issue-select
by Shubham91999
Grade a candidate open-source issue against a written rubric and decide whether it is worth taking as a first contribution. Use when evaluating a GitHub issue URL or an eval snapshot file as a potential first issue.
triage
by surreptakos
Move issues and external PRs through a state machine of triage roles, categorise, verify, grill if needed, and write agent-ready briefs.
atlassian-cli
by hoopit
Use when working with Jira or Confluence from command line, including authentication, searching issues with JQL, bulk operations, sprint reports, or creating/updating work items using acli
google-typescript-style
by chughtapan
Use when writing or reviewing TypeScript under Google's source, type-system, naming, module, documentation, and conformance conventions. Do not use for plain JavaScript; use google-javascript-style instead.
project-review
by JeriC4o
"Whole-codebase review on the current branch (or branch given as argument). Reads all source files and done plans, runs fix loop and self-review loop until APPROVE."
dailybot-tasks
by DailybotHQ
Manage Dailybot Plan via the CLI — boards, columns, tasks, projects, goals and milestones. Read the workspace in one call (pulse, what needs attention, recent activity, goal progress), poll what changed since a cursor, create/update/move tasks and set their owner, comment with @mentions, relate, attach files, watch or mute, run bulk operations with a server-side dry run, archive safely with a previewed consequence, administer boards (columns, members, saved views), post project updates so the team sees what an agent did, and work a task you were handed (read the whole card with task brief, write back attributed to the agent). Use when the developer mentions tasks, a board, a backlog, a sprint, a kanban column, a project update, a milestone or a goal, or asks what is open / overdue / blocked. Not for check-in responses (use dailybot-checkin) or form submissions (use dailybot-forms).
反ai-skill
by Wholiver
企业级代码复杂度增强工具包。将干净、可读的代码转化为专业混淆的、为保住工作而优化的实现，展示深厚的工程成熟度和架构 sophistication。
grug-nuclear
by nuoyadev
Nuclear Grug mode. 80-90% token compression. Simple good. Complex die.
pull-request-review
by dekopon-agents
Use when asked to review, summarize, or comment on a pull request. Covers what to read before commenting, how to rank findings by risk, and the one-comment rule.
coding-standards
by dhkong01
Universal coding standards, best practices, and patterns for TypeScript, JavaScript, React, and Node.js development.
linkedin-solopreneur-skill
by ahsanishaq721
Personal Brand & AI Agency Strategist for n8n Workflow Experts. Powered by Justin Welsh's "Niche of One" philosophy. Activate with /post [topic], /comment [content], /outreach [person/goal], /profile [section], /ideate, or /newsletter [topic]. Use this skill whenever the user wants to write LinkedIn posts, comments, outreach messages, newsletter content, or build their personal brand as an AI automation expert.
evidence
by nitedani
"Showing a behavior to a reader who wasn't there: browser control, reproducing it in the running app as a person would, a screenshot, a video or the request, and uploading it. Shared by open-issue and pull-request."
huluhuluu-blog-style
by huluhuluu
Write Chinese-first technical blog posts in huluhuluu's personal blog style. Use when drafting or rewriting Hugo blog content under this repo, especially for tool notes, environment setup guides, tutorial series index pages, paper reading notes, source code analysis, and engineering record posts that should match the existing content/post writing style.
cocos-vscene
by jerikchan
Use when working with Cocos Creator scenes and needing to query, browse, analyze, map, or optimize scene node structures. Triggers on "vscene" commands or Chinese equivalents: 拉取场景索引, 读取场景, 分析场景, 审查场景, 场景地图, 优化场景. Also triggers on scene node lookup, scene quality scoring, scene hierarchy review, or scene structure mapping needs. Keywords: vscene, scene index, scene cache, node lookup, Cocos Creator, scene analysis, quality score, scene map, node annotation, hierarchy optimization, progressive loading
jira-comment
by yotamleo-org
Add a comment to a Jira issue ("comment on HIMMEL-46 saying ..."). Needs explicit Jira key. Not PR/GitHub comments.
triage
by SjB
Move issues and external PRs through a state machine of triage roles, categorise, verify, grill if needed, and write agent-ready briefs.
video-avatars-and-virtual-backgrounds
by tippyentertainment
Build and debug real-time video effects for conferencing and streaming: selfie segmentation virtual backgrounds, reverse segmentation, face mesh tracking, and avatar replacement using WebRTC, MediaPipe, and WebGL/WebGPU.
dispatch
by JustAzul
Dispatch work to another AI coding agent (Codex today) through agentcli: second opinion, independent diff review, cross-check, expert-persona consult, delegated subtask, follow-ups, background jobs. Use whenever the user asks Codex, GPT or OpenAI to weigh in, review or do something, even before any code is attached. Trigger: "pergunta pro codex", "manda pro codex", "o que o codex acha", "ask codex", "codex review", "get a second opinion from GPT", "act as a senior X — use codex", "cross-check this", "delegate to codex", "agentcli job", "follow up with the agent". Not for work Claude can just do (coding), in-session PR/MR review (cr), pre-deploy checks (ship-audit), risk analysis (premortem).
deep-learning
by hchouTW
"Use for any PyTorch task, even a quick question or a one-line error - load it before answering from memory: writing, debugging, reviewing or explaining nn.Module models, training/eval loops, DataLoaders, losses, optimizers, mixed precision, gradient accumulation, checkpoint/resume, DDP/FSDP, determinism, inference, speed/memory; shape/device/dtype errors, NaN or exploding loss, slow dataloaders; Transformers, RNN/LSTM, VAE/GAN/diffusion, transfer learning, LoRA/quantization, autograd/hooks, export (ONNX/torch.compile). Also ML engineering decisions: architecture and scaling, parallelism strategy, compute/cost budgets, ablations and seed noise, dataset splits and leakage, eval design, serving capacity, drift, calibration (ECE), distribution shift, physics-informed/equivariant ML for HEP. PyTorch only, not TensorFlow/JAX/scikit-learn."