LazySkills

Use when generating, preparing, monitoring, downloading, or publishing LALACHAN Xiaoyunque videos through a logged-in Chrome/CDP web UI, especially requests mentioning Lala Xia, Aya Chan, Sasa Kun, Xiaoyunque, XYQ, Seedance, 沉浸式短片, Agent 模式, reference images, reference video, Nutstore copying, low-credit generation, or avoiding the Xiaoyunque API.

Use after generating, editing, exporting, or selecting an image when the user wants it opened or local workflow defaults require a persistent visible preview; preserves the image at a stable project path and launches a detached desktop viewer that survives the agent command.

Use when opening, controlling, testing, or refining a locally developed webapp through a dedicated Xvfb/noVNC Chrome profile, especially when the user wants visible browser operation, direct DOM manipulation instead of API shortcuts, stable repeatable controls, screenshots, long-running chat workflows, or iterative fixes until the webapp completes its task.

Use when launching, controlling, documenting, debugging, or cleaning up remote Linux GUI apps through a clean Xvfb desktop, x11vnc, noVNC/websockify, browser app windows, SSH-local ports, or camera/GUI process isolation. Applies to LabVIEW, KiCad, JLCEDA, FreeCAD, Blender GUI, vendor camera tools, license activation flows, and other agent-controlled desktop apps.

Use when designing, revising, validating, documenting, or rendering mechanical CAD parts with Shapr3D .shapr archives, STEP/Parasolid references, OpenSCAD, CadQuery/build123d/OCP, FreeCAD, Blender, STEP/STL/DXF/SVG/PDF exports, print-fit compensation, versioned artifacts, exact B-rep regeneration, optical holders, C-mount/OpenHI threads, PCB/sensor holders, or old STEP reference measurements.

Use when converting scanned PDFs, image-heavy books, technical textbooks, music books, OCR output, or rough Markdown into corrected, publishable Markdown/TeX/PDF while preserving figures, equations, diagrams, tables, captions, source structure, and evidence-based validation.

Use when creating PocketPolyglot/LinguaLeaf-style multilingual pocket books from bilingual or trilingual sources, including aligned JSON, ruby/furigana/pinyin, grammar colors, bidirectional main/comment editions, color and black-white PDFs, covers, queues, and resumable long-running generation.

Use when automating WeChat chatops for LabCanvas or AgInTi through an isolated Linux GUI, direct xwechat_files database reads, private decrypt workflows, SQLite mirrors, media sync, and split fast-chat/worker-agent operations without exposing secrets or chat logs.

Use when designing, operating, repairing, or cleanly rewriting a high-quality long-form novel workflow that writes slowly by paragraph, preserves author-written source authority, separates novelist/whole-story critic/prose critic/translator sessions, leaves narrative space, rejects dense AI-like or lifeless prose, resumes durably, or publishes clean chapter PDFs.

Diagnose, privately back up, compact, verify, and hand off stalled or extremely slow Codex CLI sessions through the official app-server protocol. Use when a Codex resume/thread appears frozen, cold-loads a huge JSONL history, has excessive context or inline-image payloads, needs safe context compaction, or must preserve unfinished work without rewriting rollout JSONL or SQLite state.

Use when publishing videos through LazyEdit, AutoPubMonitor, a remote AutoPublish host, Shipinhao, YouTube, Instagram, or LALACHAN-generated videos; covers direct CLI/API publishing, current-run reuse, one-shot settings overrides, subtitle correction prompts, Nutstore AutoPublish import, and monitoring/debugging the distributed publish workflow.

Use when installing, activating, launching, documenting, or wiring JLCEDA/LCEDA Pro with MCP bridge automation for PCB design agents, including local Electron CDP launch, private activation handling, jlc-bridge extension setup, gateway checks, and MCP tools/list or pcb_ping validation.

Use when preparing, auditing, fixing, documenting, or publishing a Fun Lazying Art website media item for Musia, including ASR/STT-corrected lyrics, multilingual lyric JSON, per-vocal lyric sets, timing and word highlighting, pinyin/furigana/Jyutping ruby, 16:9 covers, manifests, website validation, and player recording.

Use when generating, correcting, reviewing, documenting, publishing, or handing off original music in Musia, including idea-to-song, lyrics-to-song, melody/chord-controlled songs, ACE-Step/YuE/SoulX route selection, vocal quality checks, song review reports, Fun Lazying Art website publishing, per-vocal lyric sets, and LALACHAN song-first video handoffs.

Use when searching or opening LibGen book search/detail pages through Chrome/CDP, selecting best book candidates by exact metadata and PDF > EPUB > other format preference, and preventing ad or Trip.com redirects with request interception.

Use when a LALACHAN story needs repeated critique and revision cycles, especially 10-round polish, shareability repair, natural Chinese dialogue, Xiaoyunque story readiness, or running a Codex exec loop with gpt-5.5 xhigh.

Use when writing, revising, criticizing, or preparing LALACHAN/RaraXia/AyaChan/SasaKun stories, Xiaoyunque prompts, dialogue, scripts, or video story drafts, especially when the user says the story sounds strange, AI-like, unsmooth, too abstract, overpatched, hard to understand, or asks for a critic/quality pass before video generation.

Use when creating a music-video handoff from a reviewed Musia song to LALACHAN/Xiaoyunque, including full-song MV planning, chorus or climax MV cuts, uploaded asset prompts, song-locked audio replacement, and LazyEdit-ready outputs.

Use when working on Musia, AI song localization, song-to-Chinese or cross-language re-singing, extracting bass/drums/vocals/other stems, lyrics, beats, chords, phrase timing, singable lyric adaptation, YingMusic-Singer-Plus or SoulX-Singer synthesis preparation, creating same-music translated singing artifacts, or fixing Fun Lazying Art per-vocal lyric/timing JSON.

Use when preparing, checking, submitting, documenting, or troubleshooting JLCPCB/JiaLiChuang PCB orders from Gerber ZIPs, including China/global site workflows, desktop assistant handoff, Chrome CDP automation, private config, OSP/surface-finish limits, quality-compensation fees, courier defaults, and review-before-payment submission.

Use when installing or using KiCad with MCP/agent automation to inspect existing PCB projects, generate or edit KiCad boards, save component datasets, validate DRC, export Gerbers/STEP, and render board previews.

Publish a local git repository to GitHub from Codex or AgInTi, including profile-style multilingual README generation, LazyingArt banner/support panels, CITATION.cff repository citation, .github/FUNDING.yml, safe commit discipline, remote creation, push, homepage, description, and topics.

Use when preparing, checking, submitting-to-payment, documenting, or troubleshooting Wenext / 未来工场 3D-print orders from STL, STP, or STEP files, including global and China site workflows, Chrome CDP automation, cart/checkout delayed loads, invoice 普票 setup, private config, and cashier/payment-boundary stops.

Calibrate and control a dual LED Arduino UNO modulation setup using YYNMOS MOS switch modules and a Waveshare INA219 Current/Power Monitor HAT. Use when working on DualLampHI-style wiring, I2C monitor detection, PWM upload, CSV/PNG power capture, calibrated LUT constant-power crossfades, flicker reduction, or interpreting per-LED voltage/current/power telemetry.

Use when setting up, operating, debugging, or documenting the AMD Kria KV260 Prophesee Metavision lab, including PetaLinux desktop launchers, custom event viewer, native metavision_viewer behavior, headless recording API, Windows X11 control center, file transfer GUI, and event-camera recovery workflows.

Coordinate the KV260 Prophesee event-camera board with the Windows host and its USB Arduino light controller. Use when Codex needs to inspect or operate the KV260 recording API, communicate with Windows 192.168.1.166 / CSG1175-P, reason about Arduino COM-port control, clone/use the four related Windows repos, or document/diagnose port conflicts, /dev/video0 ownership, firewall, phase-sync, and recording workflows.

Coordinate multiple agent sessions across machines, repos, tools, and runtimes using safe handoff packets, peer identity maps, private session mirrors, status probes, and action contracts. Use for AgentMesh-style collaboration, AgInTi agent collaboration, Codex-to-Codex handoff, board/workstation control, cross-repo orchestration, and preserving durable context without committing private logs.

Use when replacing or covering one or more detected video faces with supplied or generated image assets, including single-avatar overlays and multi-person animal masks such as panda, red panda, horse, cow, and giraffe, using accurate detection, identity tracking, background removal, alpha compositing, and ffmpeg audio-preserving output.

Use when transcribing a video, deriving topic-based section boundaries from the transcript, and splitting the video into named clips with ffmpeg plus a manifest, instead of cutting by fixed duration.

Use when transcribing audio or video into timestamped Markdown with optional WhisperX speaker diarization, selecting large-v2 or medium fallback models, and turning transcripts or chat/source notes into neutral third-person Markdown, LaTeX, and PDF reports without exposing private source context.

Use when preparing, packaging, testing, publishing, or troubleshooting npm packages, including scoped packages, CLI bins, npm 2FA/web login, env-file tokens, GitHub Actions trusted publishing, provenance, and post-publish install verification.