LIghtJUNction

Use this skill when working with the lightflow.control_if workflow that selects between two JSON values.

Use this skill when working with the lightflow.control_merge workflow that merges two JSON values.

Use this skill when working with the lightflow.control_split workflow that splits JSON arrays or objects.

Use this skill when working with the lightflow.control_switch workflow that selects JSON values by key.

Use this skill when working with the lightflow.image_crop workflow that crops PNG image artifacts.

Use this skill when working with the lightflow.image_edit workflow that creates deterministic preview image edits.

Use this skill when working with the lightflow.image_inpaint workflow that creates deterministic preview inpaint images.

Use this skill when working with the lightflow.image_invert standard workflow, inverting PNG outputs, or composing image inversion after another LightFlow image workflow.

Use for canonical, bound XRY progress, freeze, cleanup, or archive control through LightFlow.

Use for the complete canonical XRY production chain for one frozen task subject through LightFlow.

This skill guides agents through canonical preview and explicitly confirmed XRY privacy redaction for one frozen task subject.

This skill should be used when the user asks to run or change LightFlow automatic video editing with lightflow.video_auto_edit or lightflow.video_auto_edit_plan.

This skill should be used when the user asks to "configure Video Work API MCP", "connect LightFlow to Video Work API", "secure the Video Work MCP endpoint", or "troubleshoot LIGHTFLOW_VIDEO_WORK_API_URL".

Use this skill to run or modify the end-to-end lightflow.video_auto_edit package-owned workflow that plans and renders videos.

Use this skill when creating, running, or modifying the lightflow.video_auto_edit_plan workflow for automated video editing plans.

Compose a non-black, account-specific cover from a verified vehicle video frame.

Develop, review, test, and package the OniOnline multiplayer framework, SDK, native adapter, mod synchronization, and built-in one-planet-per-player mode. Use for OniOnline code or delivery work; not for ordinary ONI gameplay advice or unrelated CycleTrim…

开发或接入基于 OniOnline.Framework 的第三方联机玩法、认证消息和模组状态频道。用于玩法程序集与 SDK 集成；不用于 OniOnline 核心、原生 UI、普通缺氧游玩或 Steam 发布。

Use when reading, polling, or sending Bilibili live-room comments/danmu. Uses public APIs for reading when possible and .env BILI_COOKIE for authenticated send.

Control Oxygen Not Included through Oni MCP, including authorized autonomous play, bounded event-driven waits, colony inspection, construction and virtual-file edits. Reuse discovered interfaces and overlap static research with bounded gameplay; use a…

为缺氧（Oxygen Not Included）提供开局、制氧、养殖、温控、喷孔、电网与太空模块的选型和故障排查建议。按本体、眼冒金星及游戏版本区分机制；不授予游戏控制或源码修改权限。

当用户要求检查 ONI 复制人状态、怀疑复制人卡住/被困/不可达、报告复制人在错误地点发呆或睡觉、要求救援分诊，或需要快速复制人健康/导航审计时使用。指导 agent 先用 world_editor command=read path=/active/dupes/index.oni，只检查被标记区域，除非用户明确授权，否则不要下达救援动作。

当用户要求快速总结 ONI 附近环境/资源、开局区域概览、周围资源、附近危险、扩张方向，或问“我周围有什么”时使用。指导 agent 使用低 token 的公开只读工具，如 world_editor 虚拟文件 /active/map/viewport.md 和 /active/colony/status.oni，并返回简洁、可执行的环境/资源摘要，不执行游戏动作。

Use when iterating on the ONI MCP server so an agent can build with onim, deploy with onim dev, automatically launch/load a save, run batched runtime tests with a tester agent, and improve tool ergonomics toward long autonomous play.

Convert existing tasks into actionable, dependency-ordered GitHub issues for the feature based on available design artifacts.

Perform a non-destructive cross-artifact consistency and quality analysis across spec.md, plan.md, and tasks.md after task generation.

Generate a custom checklist for the current feature based on user requirements.

Identify underspecified areas in the current feature spec by asking up to 5 highly targeted clarification questions and encoding answers back into the spec.

Create or update the project constitution from interactive or provided principle inputs.

Assess the current codebase against the feature's spec, plan, and tasks, then append any remaining unbuilt work as new tasks to tasks.md so implement can complete it.

Execute the implementation plan by processing and executing all tasks defined in tasks.md

Execute the implementation planning workflow using the plan template to generate design artifacts.

AgentMail mailbox workflow for the assistant email lightjunction@agentmail.to. Use when reading, sending, configuring, or documenting the assistant's AgentMail email address, mailbox identity, CLI usage, or email safety rules.

Bilibili social workflow for the assistant. Use when reading Bilibili comments, posting Bilibili comments or replies, or reading/sending Bilibili private messages with the user's cookie-backed session.

lmm.best personal static website product workflow. Use this skill when the user mentions lmm.best, /username pages, selling personal websites, static profile pages, customer intake, or deploying profile websites under lmm.best/<username>. Do not confuse this…

Low-star open-source contribution workflow. Use this when preparing or doing small, low-risk PRs/issues, candidate queues, or maintainer follow-up for early-stage active repositories.

File handoff workflow using share.lmm.best. Use this skill when the user asks to deliver files, share downloads, send work products to customers, create a file link, DM a finished file, or manage the lmm.best file transfer service.

Meta workflow for improving the assistant's own skills. Use when repeated work, recurring mistakes, copied prompts, repeated command sequences, or stable operating rules should be compressed into a reusable skill instead of being rediscovered in every session.

Use for scheduled or user-triggered maintenance of the owner's own GitHub projects: triaging notifications, reading issues/mentions, checking Actions and releases, making small fixes, and leaving clean handoff notes.

Debug and verify MagicNet on a real Android root device. Use when working on MagicNet MCP usage, adb/root validation, TUN/eBPF transparent status, DNS leak diagnosis, sing-box behavior, certificate-less packet capture, eCapture/tcpdump checks, support…

Maintain MagicNet's shell KAM framework under src/MagicNet/lib/kamfw, including helpers, imports, i18n, logging, install filters, dependency checks, and EXIT handlers.

Orchestrates Android development tasks including project creation, deployment, SDK management, and environment diagnostics using the `android` command-line tool.

Translates natural language data intents into syntactically valid Perfetto SQL queries and executes them against a local trace file. Use this skill to extract slice, thread, or memory data from Android Perfetto traces using trace_processor.

Analyzes Perfetto traces to find the root cause of latency, memory, or jank issues in Android apps. Use when the user provides a Perfetto trace file and asks any question, ongoing investigation, or open-ended request to analyze its contents.

Run and interpret the CortexFS agent harness contract suite when validating protocol, authorization, ownership, cancellation or channel regressions. Use docs/evaluation.md for the separate live-model evaluation path.

Use when changing CortexFS performance, latency, throughput, memory, token usage, cost, caching, process reuse, streaming, or when comparing CortexFS with another agent runtime.

在 CortexFS 仓库中继续去重、减少重复逻辑、合并等价代码、减少代码量或函数数量、恢复上次 DRY 重构现场，或在新增代码前避免写出重复 helper 时使用。

This skill should be used when the user asks to "build a CortexFS tool", "create a CortexFS agent", "package a CortexFS extension", "write a cortexfs.object/v1 or cortexfs.object/v2 manifest", "install an external tool", or "install an external agent".

Use this skill when working in the CortexFS repository and the user asks to test CortexFS, test FUSE, test /ctx, test ctx, test agent.sh, run live tests, use Ollama or smollm2:135m, verify file/socket ABI behavior, check tests/mounts/cortexfs, inspect the…

Use this skill when working with the lightflow.flux_image_edit workflow, prompt-guided image editing, or FLUX image-edit model sync.

Use this skill when working with the lightflow.flux_inpaint workflow, masks, local repainting, or FLUX inpaint model sync.

Use this skill when working with the lightflow.flux_preview_text_to_image deterministic preview workflow or testing FLUX pipelines without a model backend.

Use this skill when working with the lightflow.flux_text_to_image workflow, syncing FLUX models, or generating images from prompts.

Use this skill when working with the lightflow.flux_text_to_image_router workflow or routing image generation between real FLUX and preview runtimes.

Add and maintain safe paired source folding comments when creating or modifying code. Use when working on source files that can benefit from compact, recoverable source views.

Archive verbose context locally, write reversible summaries, browse originals, and resolve archived references before tools execute. Use when long tool outputs or historical observations consume working context.

This skill should be used when the user asks to "投递简历", "申请岗位", "发送在线简历", "先聊再投", "apply to a job", "submit a resume", or "send an online resume".

This skill should be used when the user asks to "筛选简历", "筛选匹配岗位", "评估简历与职位匹配", "简历匹配职位", "screen a resume", "screen matching jobs", or "evaluate resume-job fit".

Add HumanGrant account-to-agent authorization to an existing project. Use when the user asks for HumanGrant, Continue with human beings, or AI sign-in with human permission. Inspect real code, keep current human login, implement scoped AI sessions, and verify…

Obtain and use limited HumanGrant permission as a running AI client. Use when a service offers Continue with human beings or a HumanGrant Markdown status page and a task needs account-authorized access. Keep the AI key local, request minimum scope, contact…

Safely draft, price, and publish open-source bounties through the api.lmm.best Open-source bounties MCP service. Use when the user wants to create or update a bounty draft, publish a bounty project, choose a gross price or number of fixes, inspect platform…

Verify and settle api.lmm.best open-source bounty submissions using real repository, GitHub, CI, deployment, test, and MCP evidence. Use when the user asks to review, accept, approve, pay, reject, close, rate, merge, or dispute a bounty challenge or its…