All authors
Claude Skills by 8-Sync-Dev
github.com/8-Sync-Dev51 skills0 installs0 views
- AutopilotUnattended backlog runner, level 5 — /autopilot. Picks items from labelled GitHub issues, STATE.md Next, or unguarded failure lines, runs each as /ship-loop (no release), batches green PRs, then ONE Ship question; writes a brief to STATE.md. Owner-only; also headless via claude -p.Votes: 0GitHub stars: 4
- ConsolidateDọn memory của repo — gộp bài học trùng, bỏ dòng sai/lỗi thời/nhiễu môi trường, chuyển quy trình đã validated sang PLAYBOOKS, rút gọn STATE. Dùng khi user gõ /consolidate, khi KNOWLEDGE.md dài hoặc digest đầu phiên toàn chuyện cũ.Votes: 0GitHub stars: 4
- DoQuick small change, level 1 — /do <việc>, "sửa nhanh", "đổi nhỏ", typo, rename, one-function fix. One pass, no plan, no commit, light checks (lint + touched test module), then report. Refuses work spanning >3 files or touching auth/secrets/deploy (use /ship).Votes: 0GitHub stars: 4
- HardenAudit tư thế bảo mật Claude Code của máy này rồi đề xuất cách siết — permission mode, tắt bypassPermissions, sandbox, deny rule, MCP server/marketplace tin cậy, hooks, cách login (claude.ai org vs API key), Channels, quyền file .env/key. Dùng khi user gõ /harden, setup máy mới, hoặc muốn cấu hình org Team an toàn nhất.Votes: 0GitHub stars: 4
- PlanningLập plan cho một task/feature trước khi code — goal, acceptance criteria kiểm được, chia task theo file, rủi ro, câu hỏi gom một lần. Không sửa code. Dùng khi user gõ /planning <task> hoặc bước Plan của /ship.Votes: 0GitHub stars: 4
- RememberGhi MỘT bài học có bằng chứng vào memory của repo (su-code/KNOWLEDGE.md) để các session sau đọc được qua digest SessionStart. Dùng khi user gõ /remember <bài học>, hoặc khi vừa có failure đã tìm ra gốc, cách làm vừa được verify, hay quyết định thiết kế cần giữ.Votes: 0GitHub stars: 4
- RulesLuật Anthropic + quy tắc 8 Sync — trả lời "X có được phép không?" hoặc kiểm tra một hành động/diff dự định làm theo Usage Policy (AUP), Consumer/Commercial Terms, Claude Code legal & compliance và quy tắc nội bộ. Dùng khi user gõ /rules <câu hỏi hoặc việc định làm>, hỏi về token subscription, tool bên thứ ba, scrape, chatbot, dữ liệu cá nhân, hoặc cần trích dẫn điều khoản.Votes: 0GitHub stars: 4
- SecQuét bảo mật diff hoặc repo hiện tại — secret (không in giá trị), file nhạy cảm đang track/trong history, dependency audit, guard/hook/deny rule của kit, rồi review code đổi theo OWASP. Dùng khi user gõ /sec, trước commit/PR/release, hoặc sau khi diff đụng auth, input, file credential, dependency.Votes: 0GitHub stars: 4
- Ship AutoAuto ship through release, level 3 — /ship-auto <việc>. Runs the full /ship process (plan → code → test → UAT → review → commit) then ONE "Ship" AskUserQuestion, then PR → CI → merge → tag/release → install → monitor until green. Owner-only.Votes: 0GitHub stars: 4
- Ship LoopGoal loop until acceptance criteria pass, level 4 — /ship-loop <mục tiêu>. Turns the goal into criteria each with a check command, then iterates implement → check → fix under Claude Code /goal with iteration/time/stuck caps; evidence in LOOP.md; optional release via one Ship question. Owner-only.Votes: 0GitHub stars: 4
- Ship TickOne iteration of an OPEN loop the owner already started (/ship-loop, /autopilot, /ship-auto, /ship --release) — /ship-tick [slug]. Model-invocable on purpose so a Claude Code scheduled task (CronCreate "*/5 * * * *") can re-enter the loop whenever the session goes idle (the /loop + loop-me pattern): standup, pick the red criterion or next step, research when stuck, implement the smallest step, check with evidence, tick; on done/stop it closes the loop and deletes its own cron. Never starts a ...Votes: 0GitHub stars: 4
- ShipChạy trọn quy trình một feature/bug như một team — plan → R&D → code → test → UAT → review/QC → docs+memory → commit, rồi release/deploy + monitor khi owner thêm cờ --release/--deploy. Dùng khi user gõ /ship <mô tả task>.Votes: 0GitHub stars: 4
- SprintNhịp scrum cho task ≥4 file hoặc ≥2 nhóm việc độc lập — chọn role theo cổng quy mô, giao theo đợt, standup (done/doing/blocked, backlog, bug, phòng khác), review, retro. Do /ship, /ship-auto, /ship-loop, /autopilot gọi; việc nhỏ dùng /do.Votes: 0GitHub stars: 4
- SysdesignSystem design for a real change (new service, storage, queue, cache, scale, AI feature) using The System Design Primer as the reading source — pick the sections, run the 4-step method with back-of-envelope numbers, write the ADR. Use when a task adds a component, changes data flow or must survive more load.Votes: 0GitHub stars: 4
- Ui UxUI/UX và giao diện (design, redesign, frontend, landing, dashboard, component, a11y, responsive, animation, theme, UX copy) — điểm vào dễ gõ `/ui-ux <việc>`; chọn đúng skill chuyên sâu (impeccable là design system chuẩn của 8 Sync) rồi làm tới khi verify trên trình duyệt.Votes: 0GitHub stars: 4
- CommandsScaffold a new Agent Skill from a plain-English description — decide whether a skill is even warranted, pick the trigger wording, then generate a spec-valid SKILL.md via `8sync skill new` and fill in the body.Votes: 0GitHub stars: 4
- 8sync CliUse in every session inside a repo with 8sync (AGENTS.md or CLAUDE.md mentions 8sync, or a su-code/ folder exists) — under omp or Claude Code. Lists every `8sync` verb (sessions, ship, find, note, run, shot/diff-img/pdf-img, gen-img, locate, harness, skill, setup/up/doctor, hz, bt, sec, lcd, clean, forge, ccc, codex-web) and the su-code/ memory layout, so the AI uses the 8sync command instead of raw shell or guessing.Votes: 0GitHub stars: 4
- Ai Microservice DesignUse when designing or adding an AI feature to a service — an LLM gateway, RAG retrieval, an agent/tool-use loop, per-member keys + usage metering, LLM tracing, or an eval gate. Produces a service topology (mermaid), an ownership table, a provider-abstraction interface, and a defined eval threshold before any model change ships. Grounded in canonical AI-gateway/RAG/agent/eval patterns; extends them with an opinionated design workflow driven by omp-native tooling (codegraph · cbm · serena · eng...Votes: 0GitHub stars: 4
- Assp SkillBrand DNA + ASSP validation framework for 8 Sync Dev (Avatar → Brand Voice → Hero Mechanism → Money Model → Offer Architect). MUST be consulted for any user-facing copy (UI microcopy, landing/pricing pages, emails, error messages): Alex Dev voice (terse, direct, dev-to-dev Vietnamese, EN tech terms kept, banned-phrase list), brand identity + standard footer, offer/pricing structure (4-week billing). Also governs validate-before-build for new product features.Votes: 0GitHub stars: 4
- Aup AuditKiểm tra repo theo Usage Policy của Anthropic (anthropic.com/legal/aup) và quy tắc 8 Sync — secret và file nhạy cảm (cả git history), dùng token Claude sai đường, tính năng rủi ro (scrape, nhắn tin hàng loạt, nhận diện khuôn mặt, chatbot thiếu disclosure, quyết định high-risk). Dùng khi user gõ /aup-audit, trước khi public repo, trước release, hoặc khi diff thêm tính năng thu thập dữ liệu / nhắn tin người ngoài.Votes: 0GitHub stars: 4
- Branch SyncCheck, deep-preview, merge to main, and sync all git branches cleanly with zero conflicts. Use when checking branch status, previewing unmerged branches, merging feature branches into main, or updating all local/remote branches to match latest main safely.Votes: 0GitHub stars: 4
- Code Review And QualityConducts multi-axis code review. Use before merging any change. Use when reviewing code written by yourself, another agent, or a human. Use when you need to assess code quality across multiple dimensions before it enters the main branch.Votes: 0GitHub stars: 4
- CodegraphUse this skill on EVERY question that touches code in a project where `.codegraph/` exists. Triggers — "where is X defined", "what calls Y", "find references to Z", "explore <subsystem>", "understand the dependency graph", "outline this file", any keyword/symbol search. Saves ~35% tokens and ~70% tool calls vs raw grep/Read/Glob. The AI MUST prefer `codegraph` commands over `rg`/`fd`/`grep`/`Read`-whole-file whenever a semantic answer is wanted.Votes: 0GitHub stars: 4
- Deep ResearchSource-heavy investigation of a topic or codebase that ends in a cited brief. Use when the user asks for deep research, a comparison, a multi-source investigation, an in-depth technical report, or "R&D to pick the best option". Token-lean - snippets before pages, compressed subagent returns, one artifact.Votes: 0GitHub stars: 4
- Encore DeployDeploy/secrets/rollout runbook for the 8syncdev Encore Cloud backend and Vercel frontends. Triggers — "deploy gate/core to encore", "push backend to encore cloud", "set encore secrets", "rollout failed on encore", "build failed on encore", "payos secrets", "rotate a secret", "deploy frontend to vercel", "production deploy". Contains proven gotchas (secrets need a redeploy; vercel drops [...slug] dirs) — read BEFORE any backend deploy, secret rotation, or prod FE deploy.Votes: 0GitHub stars: 4
- Encore Eino GoUse when building or editing a Go backend that combines Encore.dev (infrastructure-as-code services, typed SQL, cron, pubsub, auth, secrets) with CloudWeEino (LLM orchestration — ChatModel, ReAct agent, tools, retriever, streaming). Produces Encore service scaffolding, an Eino pipeline hosted behind an `//encore:api`, a green `encore test`/`go test ./...`, and a code-gen-clean `encore.gen.go`. Grounded in Encore.go v1.57.x + Eino v0.9.x canonical bases; extends with the 8syncdev go-kit workfl...Votes: 0GitHub stars: 4
- FeatureQuản lý feature LỚN nhiều phase, nhiều ngày, xuyên session (>10 file, nhiều milestone) theo nguyên lý GSD. State sống ở su-code/planning/<slug>/ (PROJECT/REQUIREMENTS/ROADMAP/STATE) + ACTIVE switch giữa nhiều feature. Việc nhỏ/bug/1-concern (≤10 file, 1 pass) dùng /auto (lái engine trực tiếp).Votes: 0GitHub stars: 4
- Full FlowUse this skill for ANY fix/dev/enhance task on the 8syncdev-pro-v2 stack that must be PROVEN on the running app — "làm full flow", "fix theo flow chuẩn", "dev + verify", "tự chạy đến done". It is the entry playbook + verification gate: ground first (codegraph), reproduce, implement DATA→BACKEND(Encore)→FRONTEND(Next), then self-verify end-to-end through BOTH the Next.js UI (oh-my-pi `browser` tool) AND Encore structured trace logs before claiming done. Composes with `codegraph` (navigate firs...Votes: 0GitHub stars: 4
- Image GenGenerate AI images, icons, logos, diagrams, illustrations, and mockups via `gpt-image-2` using `8sync gen-img` or standalone scripts. Use whenever the user or task asks to create, draw, generate, or render an image or visual asset.Votes: 0GitHub stars: 4
- Image RoutingUse this skill on EVERY read request AND before dumping large context, to decide between text and image representation. Apply whenever the AI is about to open a PDF, screenshot a URL, review a UI, inspect a long git diff, process diagrams, OR dump a big graph / call-tree / dependency map as text. First self-check your own modality (can you see images?), then consult the decision table. Picking wrong wastes tokens — but image is NOT always cheaper (see the honest economics below).Votes: 0GitHub stars: 4
- ImpeccableUse when the user wants to design, redesign, shape, critique, audit, polish, clarify, distill, harden, optimize, adapt, animate, colorize, extract, or otherwise improve a frontend interface. Covers websites, landing pages, dashboards, product UI, app shells, components, forms, settings, onboarding, and empty states. Handles UX review, visual hierarchy, information architecture, cognitive load, accessibility, performance, responsive behavior, theming, anti-patterns, typography, fonts, spacing,...Votes: 0GitHub stars: 4
- Karpathy GuidelinesUse this skill before EVERY non-trivial coding task. It enforces Andrej Karpathy-style engineering discipline — read-before-write, test-before-refactor, small steps, boring-is-better, delete-more-than-you-add. Apply whenever the user asks for code, refactor, debug, or review work; the AI MUST cite a rule from this skill before claiming "done".Votes: 0GitHub stars: 4
- Last30daysUse this skill when the user asks "what are people saying about X", "research <topic> recently", "what's trending on Reddit/X/YouTube about Y", pre-meeting/pre-call briefings, "last 30 days of Z", competitor scans, or any recency-grounded social research. It runs the `/last30days` agent skill (separately installed engine) that searches Reddit, X, YouTube, TikTok, Hacker News, Polymarket, GitHub, Bluesky and the web in parallel, scores by real engagement, and synthesizes one cited brief. Prefe...Votes: 0GitHub stars: 4
- Locate AnythingFast open-vocabulary VISUAL GROUNDING via NVIDIA LocateAnything-3B (run through the mudler/locate-anything.cpp ggml port, no Python). Turns an image + a text description into labeled bounding boxes + click-center coordinates. Use when you need EXACT PIXEL COORDINATES from an image — GUI element grounding (to click a button/field precisely), OCR/text localization, open-set object detection — rather than a prose description. Invoked as `8sync locate <image> "<target>"`. Complements a native VLM...Votes: 0GitHub stars: 4
- Nextjs AppUse when building, editing, or debugging a Next.js App Router app (Next 15/16, React 19) — anything involving the `app/` route tree, Server Components vs `'use client'`, Server Actions, Route Handlers, the fetch cache, `middleware.ts`, SEO/metadata, an Encore.ts (or other) backend client, a Turborepo monorepo, or Vercel/Docker deployment. Produces correct RSC boundaries, non-stale caching, verified UI. Grounded in the Next.js 16 App Router canon; extends with the 8syncdev repo patterns + omp-...Votes: 0GitHub stars: 4
- PonytailForces the laziest solution that actually works, simplest, shortest, most minimal. Channels a senior dev who has seen everything: question whether the task needs to exist at all (YAGNI), reach for the standard library before custom code, native platform features before dependencies, one line before fifty. Supports intensity levels: lite, full (default), ultra. Use whenever the user says "ponytail", "be lazy", "lazy mode", "simplest solution", "minimal solution", "yagni", "do less", or "shorte...Votes: 0GitHub stars: 4
- Remote ComputeRun research, replication, or benchmark code somewhere other than the bare host — a Docker sandbox, Modal serverless GPUs, or a persistent RunPod GPU pod. Use when the user picks an isolated/remote execution environment, or asks to run untrusted or GPU-heavy code safely.Votes: 0GitHub stars: 4
- Research PaperWork a paper, claim, or benchmark end to end — replicate a result, extract an implementable ML training recipe, audit a paper against its released code, draft a paper-style write-up, or run a bounded try/measure/keep-or-revert experiment loop. Use when the user wants to reproduce, verify, optimize against, or formally write up research rather than get a casual summary.Votes: 0GitHub stars: 4
- Senior FrontendFrontend development skill for React, Next.js, TypeScript, and Tailwind CSS applications. Use when building React components, optimizing Next.js performance, analyzing bundle sizes, scaffolding frontend projects, implementing accessibility, or reviewing frontend code quality.Votes: 0GitHub stars: 4
- Senior SecuritySecurity engineering toolkit for threat modeling, vulnerability analysis, secure architecture, and penetration testing. Includes STRIDE analysis, OWASP guidance, cryptography selection tables, and a map of the security scanners to reach for.Votes: 0GitHub stars: 4
- Social GrowthOPT-IN (không auto-bật). Use ONLY when the user explicitly enables this skill or asks to build social-media marketing / branding / lead-gen for 8 Sync Dev — campaigns on Facebook, YouTube, TikTok (and Fanpage / community groups), setting up a Page properly, audience insight, monthly content plan + targets/KPIs, or finding leads & customers. Composes with `assp-skill` (brand voice + avatar + offer) and `last30days` (recency research); never replaces them. Produces a concrete monthly plan with ...Votes: 0GitHub stars: 4
- Super PdfGenerate boardroom-grade PDF reports via an HTML design system + WeasyPrint — the same template family that produced the admired CloudGO review/architecture PDFs (kicker cover with chips strip + meta table, §N numbered section spine, navy comparison tables, colored callouts, pills, stat cards, running footer with page numbers). Use whenever asked for a beautiful/professional PDF, "báo cáo PDF đẹp", super pdf, review verdict PDF, architecture doc PDF, tài liệu giao sếp/khách, or "PDF như bản r...Votes: 0GitHub stars: 4
- Taste BrutalistRaw mechanical interfaces fusing Swiss typographic print with military terminal aesthetics. Rigid grids, extreme type scale contrast, utilitarian color, analog degradation effects. For data-heavy dashboards, portfolios, or editorial sites that need to feel like declassified blueprints.Votes: 0GitHub stars: 4
- Taste MinimalistClean editorial-style interfaces. Warm monochrome palette, typographic contrast, flat bento grids, muted pastels. No gradients, no heavy shadows.Votes: 0GitHub stars: 4
- Taste OutputOverrides default LLM truncation behavior. Enforces complete code generation, bans placeholder patterns, and handles token-limit splits cleanly. Apply to any task requiring exhaustive, unabridged output.Votes: 0GitHub stars: 4
- Taste RedesignUpgrades existing websites and apps to premium quality. Audits current design, identifies generic AI patterns, and applies high-end design standards without breaking functionality. Works with any CSS framework or vanilla CSS.Votes: 0GitHub stars: 4
- Taste SkillAnti-slop frontend skill for landing pages, portfolios, and redesigns. The agent reads the brief, infers the right design direction, and ships interfaces that do not look templated. Real design systems when applicable, audit-first on redesigns, strict pre-flight check.Votes: 0GitHub stars: 4
- Taste SoftTeaches the AI to design like a high-end agency. Defines the exact fonts, spacing, shadows, card structures, and animations that make a website feel expensive. Blocks all the common defaults that make AI designs look cheap or generic.Votes: 0GitHub stars: 4
- Tauri V2Use when building, migrating (v1→v2), or debugging a Tauri 2.x desktop app — adding a `#[tauri::command]`, fixing "command not allowed"/capabilities/ACL errors, wiring plugins, bundling per-OS, configuring the auto-updater, or attaching a heavy native/Python sidecar. Produces a verified Rust+webview app whose IPC round-trip and bundle target are exercised. Grounded in official Tauri v2 docs (v2.tauri.app); extends with opinionated patterns from 8syncdev's open-musik/zus/sidex and omp-native t...Votes: 0GitHub stars: 4
- Token BenchProve how much the code-intelligence stack (codegraph; codebase-memory-mcp if enabled) actually saves on a real codebase — measured token reduction vs naive grep+read, with a correctness check. Use when the user asks "how much token does codegraph save", "benchmark the token optimization", "prove >X% reduction on this repo", or wants evidence the STEP-0 stack is worth it. Runs a real codebase, not a synthetic claim.Votes: 0GitHub stars: 4