8-Sync-Dev
@8-Sync-Dev Organization
Public Skills
deep-research
by 8-Sync-Dev
Source-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.
tauri-v2
by 8-Sync-Dev
Use 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 tooling (codegraph, codebase-memory, serena, engine_*, browser).
ai-microservice-design
by 8-Sync-Dev
Use 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 · engine_* · browser · 8sync verbs).
nextjs-app
by 8-Sync-Dev
Use 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-native tooling (codegraph · codebase-memory · serena · engine_* · browser).
encore-eino-go
by 8-Sync-Dev
Use 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 workflow and omp-native code-intelligence tooling.
rules
by 8-Sync-Dev
Luật 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.
harden
by 8-Sync-Dev
Audit 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.
image-routing
by 8-Sync-Dev
Use 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).
image-gen
by 8-Sync-Dev
Generate 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.
zai-vision
by 8-Sync-Dev
FALLBACK only. Use this skill ONLY when the CURRENT model is text-only (GLM-5.2 and older). SKIP the entire skill when the session model is GLM-5.3 / GLM-5.3-Flash / any zai/glm-5.3* — those are native VLMs (https://docs.z.ai/guides/vlm/glm-5.3-flash) and MUST read images in-session. Do not call mcp__zai_vision_* from a GLM-5.3 session. For text-only models, this is the GLM-5V MCP bridge after 8sync shot / pdf-img / diff-img.
locate-anything
by 8-Sync-Dev
Fast 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 / zai-vision (which describe) and the browser tool (which clicks).
super-pdf
by 8-Sync-Dev
Generate 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 review MR".
code-review-and-quality
by 8-Sync-Dev
Conducts 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.
remote-compute
by 8-Sync-Dev
Run 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.
senior-security
by 8-Sync-Dev
Security 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.
karpathy-guidelines
by 8-Sync-Dev
Use 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".
codegraph
by 8-Sync-Dev
Use 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.
research-paper
by 8-Sync-Dev
Work 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.
social-growth
by 8-Sync-Dev
OPT-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 per-channel targets, content calendar, funnel, and lead-capture mechanics — never generic "post consistently" advice.
impeccable
by 8-Sync-Dev
Use 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, layout, alignment, color, motion, micro-interactions, UX copy, error states, edge cases, i18n, and reusable design systems or tokens. Also use for bland designs that need to become bolder or more delightful, loud designs that should become quieter, live browser iteration on UI elements, or ambitious visual effects that should feel technically extraordinary. Not for backend-only or non-UI tasks.
senior-frontend
by 8-Sync-Dev
Frontend 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.
ponytail
by 8-Sync-Dev
Forces 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 "shortest path", and whenever they complain about over-engineering, bloat, boilerplate, or unnecessary dependencies.
full-flow
by 8-Sync-Dev
Use 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 first); never replaces it.
encore-deploy
by 8-Sync-Dev
Deploy/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.