moviepilot-agent

Use this skill when the Agent needs to decide where a capability, rule, workflow, runtime anchor, script, job, document, persona change, plugin asset, or repository maintenance item should live. It provides a layered map of the Agent's durable assets so requests to “落实到 all”, “整理自身框架”, or “细化自身” do not bloat memory or scatter files.

Use this skill when the user asks to 自检自身、重新蒸馏、进化版 Agent、增强回去、避免死循环处理、能力增益评估，或在瘦身与增强之间取舍。它先盘点已处理对象、能力风险和真正增益点，再决定增强、修复、拆分、瘦身或停止，不把行数减少当成默认目标。

Use this skill when the Agent needs an executive-control layer before acting: classify user intent, decide direct execution vs read-only inspection vs button confirmation vs clarification, avoid over-asking, and keep work moving to the smallest safe completed state. Trigger especially when the user says 继续、落地、放开手脚、接管、自己判断、按方案推进，or complains about typing confirmations that should be buttons.

Use this skill when the Agent needs to read feedback signals before deciding whether to act, correct itself, distill a lesson, or ask for clarification. It treats user wording, repeated failures, tool errors, missing evidence, and activity context as sensory inputs, then routes to the right organ without overreacting.

Use this skill when the Agent needs to turn experience into a repeatable growth loop: observe signals, decide whether learning is needed, distill reusable lessons, place them into the right asset layer, verify the change, and keep future behavior lighter rather than heavier. Trigger when the user asks to 主动成长、持续进化、经验闭环、复盘后改进, or when repeated failures show a missing habit rather than a one-off bug.

Use this skill when the Agent needs to distinguish safe self-improvement from high-risk mutation. It guards credentials, destructive actions, MoviePilot core files, downloads, restarts, external service changes, and over-broad memory edits while still allowing authorized low-risk /config/agent and /config local plugin work to proceed.

Use this skill when the Agent needs to map a request or failure to one of its long-term capability organs: executive control, proactive orchestration, self-correction, growth cycle, capability map, distillation metabolism, verification, or repository maintenance. It prevents dumping everything into memory and keeps each organ's responsibility clear.

Use this skill for whole-Agent proactive orchestration before acting or replying: run the tg-button-interaction gate first, classify user intent, choose direct execution vs button choice vs read-only inspection vs clarification, run tasks to closure, trigger verification and handoff hooks, and avoid step-by-step mechanical behavior or after-the-fact apology patches.

Use this skill when the user corrects the Agent, says the Agent misunderstood, points out over-caution, mechanical behavior, wrong persona intensity, wrong boundary interpretation, or asks the Agent to turn a mistake into a durable rule. It converts correction into diagnosis, repair, and properly layered memory/skill/persona/workflow updates.

在需求不明确、需要设计取舍、创建新功能/新行为或范围较大的创造性工作之前使用；用户已给出明确最小修改、修复或落地执行指令时，不用它阻塞直接执行。

中文 review 沟通参考——话术模板、分级标注（必须修复/建议修改/仅供参考）、国内团队常见反模式应对。仅在用户显式 /chinese-code-review 时调用，不要根据上下文自动触发。

中文 commit 与 changelog 配置参考——Conventional Commits 中文适配、commitlint/husky/commitizen 中文模板、conventional-changelog 中文配置。仅在用户显式 /chinese-commit-conventions 时调用，不要根据上下文自动触发。

中文文档排版参考——中英文空格、全半角标点、术语保留、链接格式、中文文案排版指北约定。仅在用户显式 /chinese-documentation 时调用，不要根据上下文自动触发。

国内 Git 平台配置参考——Gitee、Coding.net、极狐 GitLab、CNB 的 SSH/HTTPS/凭据/CI 接入差异与镜像同步配置。仅在用户显式 /chinese-git-workflow 时调用，不要根据上下文自动触发。

Use this skill as the fallback dispatcher for unknown or plugin-provided slash commands after moviepilot-direct-routes cannot map the request. It can list available slash commands, inspect plugin command capabilities, and execute a selected command. Do not use it before moviepilot-direct-routes for exact MoviePilot commands, direct 115/magnet/ed2k links, or obvious built-in command aliases.

Use this skill when the user asks to create, modify, debug, validate, or scaffold a MoviePilot local plugin. Covers MoviePilot V2 plugin development, _PluginBase implementations, package.v2.json/package.json market metadata, plugins.v2/plugins source layout, PLUGIN_LOCAL_REPO_PATHS local plugin sources, plugin APIs, Vuetify JSON forms/pages/dashboards, Vue module federation remote components, get_render_mode, get_sidebar_nav, plugin sidebar pages, commands, services, workflow actions, agent tools, and local install/reload flows. Also use for Chinese requests mentioning 编写插件、本地插件源, 插件开发, V2插件, 插件市场, 本地安装插件, 插件热加载, 前端联邦, 侧栏入口, Vue插件页面.

当面对 2 个以上可独立只读调查、检索、诊断或规划的任务时使用；用 MoviePilot subagent 工具并发获取私有上下文，由主 Agent 综合决策。

当已有书面实现计划且需要按检查点执行、验证和收尾时使用；适用于 MoviePilot Agent 可用工具链，不依赖外部任务清单工具。

当实现完成、所有测试通过、需要决定如何集成工作时使用——通过提供合并、PR 或清理等结构化选项来引导开发工作的收尾

Use this skill when designing, scaffolding, validating, packaging, or iterating a MoviePilot V2 local plugin repository, especially when the user mentions local plugin sources, /config/FFplugin, PLUGIN_LOCAL_REPO_PATHS, package.v2.json, plugins.v2, PR 5687, plugin market metadata, Vuetify JSON plugin pages, Vue module-federation plugin pages, plugin APIs, plugin services, slash commands, workflow actions, dashboards, or agent tools.

MCP 服务器构建方法论 — 系统化构建生产级 MCP 工具，让 AI 助手连接外部能力。仅在用户显式要求 MCP server / MCP 工具开发时调用，不要抢占 MoviePilot 主业务路线。

Use this skill for advanced MoviePilot/NASTool custom identifier rulecraft when simple generate-identifiers rules are not enough: anime or TV releases with non-standard seasons, absolute episode numbering, TMDB episode groups, noisy bilingual torrent names, conflicting years/resolutions/bit-depth numbers, direct metadata binding, or anti-collision regex design.

Use this skill when maintaining the moviepilot-agent Git repository, syncing Agent capability assets, prompting for repository sync after /config/agent capability changes, removing tracked directories, fixing failed Git pushes, initializing or restoring GitHub SSH/Deploy Key access, or performing handoff-safe Git cleanup and verification.

Use this skill only when the user explicitly asks to interact with MoviePilot through REST/HTTP API endpoints, needs API endpoint details, or a required MoviePilot operation cannot be covered by existing MCP tools, direct routes, or moviepilot-cli. Do not use it as the default path for ordinary media, download, subscription, site, library, or plugin tasks.

MUST-USE when the user sends a slash command, 115 share link, magnet/ed2k link, or obvious natural-language command alias. Do not route through AI reasoning, media search, library check, or database query when a direct slash command exists. Covers /cookiecloud, /sites, /subscribes, /downloading, /transfer, /redo, /clear_cache, /restart, /version, /clear_session, /stop_agent, /skills, /site_signin, /update_covers, /p115_full_sync, /p115_inc_sync, /p115_add_share, /p115_share_strm, /ol, /p115_strm, /sh, /p115_checkin, and similar. This is the first routing skill for direct command execution; do not bypass it with general tools.

Use this skill when designing, reviewing, or improving MoviePilot-specific UI: local plugins, settings pages, dashboards, Vuetify JSON forms, Vue plugin pages, MP 运维助手界面, “参考115 STRM”, “今日UI大学习”, “太丑了”, “像正经插件”, icons, layout, micro-interactions, accessibility, and design distillation for MoviePilot.

收到代码审查反馈后、实施建议之前使用，尤其当反馈不明确或技术上有疑问时——需要技术严谨性和验证，而非敷衍附和或盲目执行。

完成任务、实现重要功能或合并前使用，用于验证工作成果是否符合要求；在 MoviePilot Agent 中优先使用只读子代理或本地验证命令，不依赖外部审查者。

MUST-USE when the user asks to search MoviePilot tracker/torrent resources for a movie, TV show, anime, or ambiguous title. Trigger for: 搜资源, 搜索资源, 找资源, 查种子, 找种子, 搜片源, 下载哪个, 资源筛选, 4K资源, 1080p资源, BluRay资源, 搜片, 搜[片名], 找[片名], 查[片名]. Resolve media through the shortest path, present one compact card with buttons first, then route to resource search, 115 search, or subscription based on user choice. This is the primary resource discovery skill; do not bypass it with direct search_media + text reply.

Use this skill when the user asks the Agent to 复盘经验、吸收知识、沉淀、去除糟粕、 焚烧炼化自身、进化出新的知识/技能/记忆，或要求把经验落实到 skills、memory、 scripts、jobs、runtime、docs/archive 或 all。该技能负责将用户反馈、纠错、 经验和长期规则分层代谢为可复用能力资产，而不是机械复制原话。

Use this skill when the Agent needs to design, review, split, merge, or maintain Agent skills and rule assets using skill-based architecture principles. Trigger for requests about 好的 skill 写法、skill 架构、规则迁移、规则散落治理、SKILL.md 过长、触发不稳定、thin shell、routing source、AAR、技能生命周期维护, or after learning from external skill architecture materials.

当在当前会话中执行包含独立任务的实现计划时使用；主 Agent 负责写入与确认，子代理仅做隔离的只读调查、审查或方案建议。

遇到任何 bug、测试失败或异常行为时使用，在提出修复方案之前执行；要求先定位根因、再做最小修复、最后用新鲜证据验证。

在实现功能、修复 bug 或改变可执行行为前使用；要求先定义失败验证或可观察验收，再做最小实现，最后用新鲜证据确认通过。

MUST-CHECK before every user-facing reply. This is the Agent's always-on interaction spine and primary UX weapon: confirmations, continue/next-step, authorize, agree/cancel, retry, media/site/season/quality/action selection, command selection, repository/scheduler/workflow decisions, and safe smoke tests must use ask_user_choice whenever the channel can support it. It should simplify every bounded interaction by choosing good defaults, offering compact action buttons, and preventing the user from typing routine confirmations. Text fallback is only temporary continuity while diagnosing broken Telegram callbacks; it is never a successful replacement for real buttons. For perfection claims, require both a live callback and log-level callback evidence.

Use this skill when you need to retry failed file transfers/organizations. Given one or more failed transfer history record IDs, this skill guides you through querying the failure details, deleting the old records, and re-identifying and re-organizing the files. Supports batch processing multiple files from the same media.

当需要开始与当前工作区隔离的功能开发或执行实现计划之前使用——创建具有智能目录选择和安全验证的隔离 git 工作树。

在开始任务时用于确认是否有适用技能，并按 MoviePilot Agent 的技能路由执行；这是技能入口纪律，不覆盖 MoviePilot 身份、用户明确指令或安全边界。

MUST-RUN before claiming any work is complete, fixed, or passing. Trigger before saying "完成", "搞定", "测试通过", "已修复", or any completion claim. Run fresh verification commands, read output, check exit codes, and only then state the conclusion. Never use memory, previous runs, or assumptions as evidence. This is a non-negotiable pre-completion gate. If you cannot show fresh evidence, you cannot claim completion.

工作完成标准流程 - 用于任务完成后的主动闭环：最小验证、整理归纳、必要归档与 hook 检查。避免机械式申请、逐步等待、手写 activity 或污染长期记忆。