Agent skill

Generate Script

by ArcReel in ArcReel/ArcReel

提示词编写:调用项目配置的文本模型,为正式脚本中待编写的分镜 / 视频单元补出 imageprompt 与 videoprompt(参考生视频改写单元正文);ad 项目尚无正式脚本时整份生成。由 create-episode-script 子智能体调用,输入是正式脚本自身的内容与 project.json。

AGPL-3.0Auto-check passedMedia & Creative

Install Generate Script

skills CLI
$ npx skills add ArcReel/ArcReel --skill generate-script -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install ArcReel/ArcReel generate-script --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/ArcReel/ArcReel.git skills-src && mkdir -p .claude/skills && cp -r skills-src/agent_runtime_profile/.claude/skills/generate-script .claude/skills/generate-script && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
generate-script
GitHub stars
5.4k
Token cost
~1.4k tokens
SKILL.md length
294 words
Files
1
Skills in repo
18
Repo updated
First seen
Licence
AGPL-3.0

At a glance

提示词编写:调用项目配置的文本模型,为正式脚本中待编写的分镜 / 视频单元补出 imageprompt 与 videoprompt(参考生视频改写单元正文);ad 项目尚无正式脚本时整份生成。由 create-episode-script 子智能体调用,输入是正式脚本自身的内容与 project.json。

  • Works in 3 steps: 项目目录下存在 project.json(含 style / overview… → 正式脚本 scripts/episode_N.json 已存在(N 是目标集的集… → 约束失败产出保留为待修复草稿,不丢弃重抽:参考生视频提示词编写的产出违反内容约束时…
  • Tasks that involve AI video generation
  • SKILL.md covers 前置条件, 用法, 生成流程 and 输出格式, plus 1 more section
  • Calls python

What it does

Generate Script is an agent skill from ArcReel/ArcReel. 提示词编写:调用项目配置的文本模型,为正式脚本中待编写的分镜 / 视频单元补出 imageprompt 与 videoprompt(参考生视频改写单元正文);ad 项目尚无正式脚本时整份生成。由 create-episode-script 子智能体调用,输入是正式脚本自身的内容与 project.json。

Its SKILL.md is about 1.4k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It sits in Media & Creative, covering AI video generation. The repository describes itself as: AI Agent 驱动的开源可自部署视频工作台:将小说与剧本转为角色、场景、道具资产、分镜、视频和剪映草稿,支持跨镜头一致性、多供应商与费用追踪 | Self-hosted AI video workspace for stories, storyboards and short-form video production. The licence is AGPL-3.0.

When your agent uses it

  • Tasks that involve AI video generation

Example prompts

  • “/generate-script”

Requirements

  • Python 3

Workflow steps

3 steps, taken from the first numbered list in SKILL.md.

  1. 项目目录下存在 project.json(含 style / overview / characters / scenes / props)
  2. 正式脚本 scripts/episode_N.json 已存在(N 是目标集的集 ID,取自计划 target.episode,不是第几集)(drama / narration /…
  3. 约束失败产出保留为待修复草稿,不丢弃重抽:参考生视频提示词编写的产出违反内容约束时,正式文件不写,产出连同逐条违约报告落到 *.invalid.json。用 open_draft 读取草稿及 revision,按 violations[] 修复完整 content,再用…

What it can do on your machine

Read from SKILL.md and the folder at commit dec992c. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Shell commands in SKILL.md call:

    • python

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Generate Script loads about 1.4k tokens when it runs. Until then it costs about 43 tokens; SKILL.md has 294 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~43
When it runs · the whole SKILL.md, loaded when a task matches
~1.4k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from ArcReel/ArcReel at commit dec992c, republished under its AGPL-3.0 licence (© ArcReel). 294 words, ~1,376 tokens.

Download SKILL.mdSave it as .claude/skills/generate-script/SKILL.md (or your agent's skills folder).
name
generate-script
description
提示词编写:调用项目配置的文本模型,为正式脚本中待编写的分镜 / 视频单元补出 image_prompt 与 video_prompt(参考生视频改写单元正文);ad 项目尚无正式脚本时整份生成。由 create-episode-script 子智能体调用,输入是正式脚本自身的内容与 project.json。
user-invocable
false

generate-script

调用项目配置的文本生成模型(Gemini / Ark / OpenAI / 自定义供应商,由 project.json 决定), 为正式脚本中待编写的条目补出视觉层。剧本里的 image_prompt / video_prompt 是后续图像 / 视频生成的"种子",Prompt 质量基本决定了画面质量——所以本 skill 是 ArcReel 整条 pipeline 中最值得重点优化的一环。

前置条件

  1. 项目目录下存在 project.json(含 style / overview / characters / scenes / props)
  2. 正式脚本 scripts/episode_N.json 已存在(N 是目标集的集 ID,取自计划 target.episode,不是第几集)(drama / narration / reference_video):内容确认即把脚本规划整集转为正式脚本,全部条目带待编写标记。确认有两条等价路径:用户在 Web 端点击确认,或在对话中明确同意后由主 Agent 调用 mcp__arcreel__confirm_script_review({"episode_id": N})。本工具只读正式脚本、不读脚本规划:脚本规划缺失或重跑后尚未确认都不影响编写。
    • ad(广告/短片):尚无正式脚本时本工具按 project.json 的 brief + products(含 selling_points)+ target_duration 整份生成(后端按审定的带货八段框架配比表构建 prompt,products 为空自动分流通用短片),brief 与 products 至少一项。结果直接成为正式脚本,不经内容确认;引用里的新角色 / 场景 / 道具由模型列进 new_assets 并给出处理决定,随正式脚本登记为待生成资产,与已登记同类资产同名的自动归并,回执列出本次新登记的资产。已有正式脚本时与其他路线一样只编写待编写条目,整份重做见下文。
  3. 约束失败产出保留为待修复草稿,不丢弃重抽:参考生视频提示词编写的产出违反内容约束时,正式文件不写,产出连同逐条违约报告落到 *.invalid.json。用 open_draft 读取草稿及 revision,按 violations[] 修复完整 content,再用 patch_draft 提交;随后以相同 episode_id 与 doc_type: reference_prompt_authoring 调 promote_draft,仍违约则继续 open → patch → promote,无轮次上限。

用法

通过 MCP 工具调用(项目名由 session 绑定,不需要传):

text
mcp__arcreel__generate_episode_script({"episode_id": N})
mcp__arcreel__generate_episode_script({"episode_id": N, "instructions": "<附加指令原文,可选,无则省略>"})
mcp__arcreel__generate_episode_script({"episode_id": N, "dry_run": true})   # 仅预览 prompt

输出路径由工具内部固定为 {project}/scripts/episode_{N}.json,不支持自定义; 如需重命名或归档,请在 Web 端操作。

补充提示词(author_prompts)

正式脚本里带待编写标记的条目(内容确认转换出的全部条目、手动新增的分镜 / 单元)还没有视觉层。 工作流计划的 next_action.type 为 author_prompts 时,requested_ids 就是这些待编写条目。 不传 entry_ids 调用即编写全部待编写条目,写回后标记清除:

text
mcp__arcreel__generate_episode_script({"episode_id": N})
  • 默认补缺:图片提示词与视频提示词各自整份判断,已有的保留,只补缺失的那一份;参考生视频按待编写 标记展开单元正文。已编写完成的条目(包括用户手写的)保持原样。
  • entry_ids 只划定范围:点名条目照样补缺,视觉层已齐的跳过,回执列出跳过项。
  • rewrite: true 是显式重写:重新编写范围内条目的全部视觉层,只在用户明确要求覆盖这几条时传入;内容字段 (台词、旁白正文、对应原文)与备注、已生成产物照常保留。会覆盖已有内容时工具先返回 prompt_overwrite_required,回执正文是服务端生成的丢失清单,params.prompt_overwrite.revision 是认可令牌: 把清单原文转述给用户,得到同意后才以该 revision 作为 overwrite_revision 重新调用。
  • 没有要编写的条目时工具不调用模型、不改剧本,回执会说明。整集重做走重跑脚本规划并重新确认。
广告/短片整份重做(regenerate)

用户要求重新生成整份广告脚本时传 regenerate: true,不需要先删正式脚本:

text
mcp__arcreel__generate_episode_script({"episode_id": N, "regenerate": true, "instructions": "<可选>"})
  • 已有正式脚本时工具先返回 script_overwrite_required,回执正文是丢失清单(将移除的条目与名下的分镜图、视频等), params.script_overwrite.revision 是认可令牌:把清单原文转述给用户,得到同意后以 regenerate: true + overwrite_revision 重新调用。
  • 上一次登记的资产在重做后保留,即使新脚本不再引用。
  • 整份生成的产出违约(引用了未登记也未列进 new_assets 的名字、处理决定解析不出、结构不合)时任务失败, 问题码 ad_script_rejected,正式脚本、资产与草稿都不变。把失败原因转述给用户,按原因补一句附加指令再生成。

重要:生成剧本必须调用上述 MCP 工具。此 skill 不提供任何 Python/Shell 脚本,不得用 BASH 调 python .../scripts/*.py。

生成流程

MCP 工具内部通过 ScriptGenerator 完成以下步骤:

  1. 加载 project.json — 读取 content_mode、characters、scenes、props、overview、style
  2. 加载正式脚本 — 在范围内(全部待编写条目,或 entry_ids 点名的条目)按补缺或显式重写选出本次编写的条目与字段;ad 尚无正式脚本或 regenerate 时整份生成
  3. 构建 Prompt — 由 lib.prompts.prompt_builders_script、lib.prompts.prompt_builders_reference 或 lib.prompts.prompt_builders_ad 生成,输入是这些条目的内容字段
  4. 调用 TextBackend — 由 TextGenerator 按项目配置选择文本模型,传入 Pydantic schema 作为 response_schema 强约束 JSON 结构
  5. Pydantic 验证与写回 — LLM 只产出视觉层,后端按条目 id 写回正式脚本,内容字段不进 LLM 输出,从工程上杜绝其经 Structured Outputs 漂移:
    • narration → NarrationVisualEpisodeScript(segment_id + image_prompt + video_prompt)
    • drama(storyboard,含 grid_storyboard)→ DramaVisualScript(scene_id + image_prompt + video_prompt)
    • ad 分镜 → AdVisualScript(shot_id + image_prompt + video_prompt);ad 整份生成 → AdEpisodeScript 加 new_assets(storyboard)或 AdReferenceFlatScript(reference_video),新增资产按处理决定改写引用后与正式脚本同一次写入登记
    • reference_video → ReferencePromptAuthoringFlatScript:待编写单元按顺序各一段改写后的正文,台词逐字保留
  6. 补充元数据 — episode(集 ID)、content_mode、novel(项目 title + 本集标题,无标题时为「第 {播出位置} 集」)、时间戳。这些字段对 LLM 隐藏(SkipJsonSchema),由后端从 project.json 注入,避免 LLM 幻觉污染下游消费方(如剪映草稿)。
    • 注:任何骨架的剧本都不写入顶层 generation_mode。生成模式是项目级事实(project.json 的 generation_mode,创建时锁定),剧本骨架种类本身即生成模式的体现;消费方一律读 project.json 分派,不得从剧本上找该字段。

输出格式

生成的 JSON 文件保存至 scripts/episode_N.json,核心结构:

  • title:LLM 写入的剧集标题
  • episode / content_mode / novel(含 title、chapter):由后端 _add_metadata 注入,不依赖 LLM 输出
  • 旁白/解说:segments[](每个分镜含 novel_text、duration_seconds、segment_break、出场角色 / 场景 / 道具 —— 内容确认时从脚本规划转入,之后在正式脚本上修改;image_prompt、video_prompt —— 由 prompt_authoring 生成)
  • 剧情演绎:scenes[](每个分镜含 image_prompt、video_prompt、duration_seconds,以及内容确认时从脚本规划转入的 utterances、source_text、characters_in_scene 等)
  • 广告/短片:shots[](每个分镜含 section、voiceover_text、products_in_shot、image_prompt、video_prompt、duration_seconds 等);总时长偏离 target_duration 超阈值仅日志提醒,不阻塞保存
  • 参考生视频:video_units[](每个视频单元含 text、duration_seconds 等)
  • metadata:created_at、updated_at、generator

条目数与全集总时长不落盘:它们逐读剧本即得,由项目摘要读时计算,落一份只会与正文漂移。

--dry-run 输出

打印将发送给文本模型的完整 prompt 文本,不调用 API、不写文件。用于检查 prompt 质量和长度。

两种生成模式(storyboard / reference_video)在 narration / drama 下的数据路径、脚本规划子智能体、schema 选择详见 .claude/references/generation-modes.md;ad 的路径见 CLAUDE.ad.md。

© ArcReel, AGPL-3.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in agent_runtime_profile/.claude/skills/generate-script of ArcReel/ArcReel.

Open the folder on GitHubat commit dec992c

Compare with similar skills

Generate Script next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Generate Script compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Generate Script this skillArcReel/ArcReel5.4k—~1.4kAutomated safety check: PassAGPL-3.0
Video Generationbytedance/deer-flow83k4 repos~1.4kAutomated safety check: PassMIT
Video Shotseternityspring/reelbench-skills8682 repos~1.8kAutomated safety check: NotesApache-2.0
Video Cover Imageitwanger/toBeBetterJavaer18k—~3.3kAutomated safety check: PassNone
Seedancesongguoxs/seedance-prompt-skill2.9k1 repos~2.5kAutomated safety check: PassNone
HyperFrames Video Entry Pointheygen-com/hyperframes59k3 repos~5.2kAutomated safety check: PassApache-2.0

Similar skills

  • Video Generation

    bytedance/deer-flow

    Generates short videos from a structured JSON prompt, optionally guided by a reference image used as the first or last frame.

    83k GitHub starsUsed in 4 repos~1.4k tokens
    Media & CreativeAuto-check passed
  • Video Shots

    eternityspring/reelbench-skills

    拉片:把一条成片拆成逐镜头的分析表——每个镜头的时长、景别、类别、运镜、画面. An agent skill from eternityspring/reelbench-skills.

    868 GitHub starsUsed in 2 repos~1.8k tokens
    Media & CreativeAuto-check: notes
  • Video Cover Image

    itwanger/toBeBetterJavaer

    Generate matched 3:4, 16:9, and 4:3 short-video cover images from toBeBetterJavaer video scripts or AI/Java technical topics.

    18k GitHub stars~3.3k tokensUpdated today
    Media & CreativeAuto-check passed
  • Seedance

    songguoxs/seedance-prompt-skill

    This skill should be used when the user asks to "generate video prompts", "create Seedance prompts", "write video descriptions", mentions "Seedance", "seedance", "即梦", "即梦平台", "视频提示词", "视频生成"…

    2.9k GitHub starsUsed in 1 repo~2.5k tokens
    Media & CreativeAuto-check passed
  • HyperFrames Video Entry Point

    heygen-com/hyperframes

    Entry point for making, editing and rendering videos from HTML compositions with HyperFrames, routing each request to the right workflow.

    59k GitHub starsUsed in 3 repos~5.2k tokens
    Media & CreativeAuto-check passed
  • Lanshu Create AI Presenter Video

    cclank/lanshu-create-ai-presenter-video

    Turn a topic or finished script into a complete, publish-ready explainer video — led by an AI presenter from an authorized adult presenter image, or performed in one of nine visual explainer styles…

    2.5k GitHub stars~3.6k tokensUpdated 2 days ago
    Media & CreativeAuto-check passed

More from ArcReel/ArcReel

All 18 skills in this repo
  • PR AI Review Loop

    ArcReel/ArcReel

    PR AI review 收敛。用户或 team-lead 要求启动或继续审查—修复循环时使用;供本地实现者或受委派的 review-looper 调用,不用于 GitHub reviewer 产出审查意见或仅处理单条评论。

    5.4k GitHub stars~667 tokensUpdated today
    Auto-check passed
  • Tech Doc Style Chinese

    ArcReel/ArcReel

    在撰写、改写、校对或审阅中文技术文档、产品文案、界面文案、Markdown 文档、API 说明、操作手册、故障排查或运维文档时使用。采用克制、准确、可扫读的中文技术写作风格;保留原文事实、限制和机器可读内容;按任务读取术语排版、API 状态文案、项目覆盖或受控中文技术写作参考。

    5.4k GitHub starsUsed in 1 repo~845 tokens
    Auto-check passed
  • Translate Docs

    ArcReel/ArcReel

    Translate every dirty ArcReel documentation source into English and refresh the translation lockfile.

    5.4k GitHub stars~1.8k tokensUpdated today
    Auto-check passed
  • Adapt Custom Endpoint

    ArcReel/ArcReel

    当用户要求把采用 JSON 提交后轮询协议的视频或图片供应商接入 ArcReel,或要求编写、验证、测试、保存自定义调用端点定义时使用。

    5.4k GitHub stars~413 tokensUpdated today
    Auto-check passed
  • Afk Team Workflow

    ArcReel/ArcReel

    把一个 Spec 的全部子 issue(或一组显式 issue)组建团队无人值守跑到全部合并或明确暂停. An agent skill from ArcReel/ArcReel.

    5.4k GitHub stars~794 tokensUpdated today
    Auto-check passed
  • Update Docs

    ArcReel/ArcReel

    根据最近的 git 改动,更新面向用户的文档(README 双语、入门教程、部署、剪映导出等)。手动调用. An agent skill from ArcReel/ArcReel.

    5.4k GitHub stars~544 tokensUpdated today
    Auto-check: notes

Questions about Generate Script

What does Generate Script do?

提示词编写:调用项目配置的文本模型,为正式脚本中待编写的分镜 / 视频单元补出 imageprompt 与 videoprompt(参考生视频改写单元正文);ad 项目尚无正式脚本时整份生成。由 create-episode-script 子智能体调用,输入是正式脚本自身的内容与 project.json。. Generate Script is an agent skill from ArcReel/ArcReel.

When should I use Generate Script?

Generate Script fits situations like: tasks that involve AI video generation.

How do I install Generate Script in Claude Code?

Run `npx skills add ArcReel/ArcReel --skill generate-script -a claude-code`. Or copy the skill folder (agent_runtime_profile/.claude/skills/generate-script in ArcReel/ArcReel) into .claude/skills/generate-script in your project. Claude Code loads it when a task matches its description.

How do I install Generate Script in Codex?

Run `npx skills add ArcReel/ArcReel --skill generate-script -a codex`. Or copy the skill folder (agent_runtime_profile/.claude/skills/generate-script in ArcReel/ArcReel) into .agents/skills/generate-script in your project. Codex loads it when a task matches its description.

Can I use Generate Script in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add ArcReel/ArcReel --skill generate-script -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/generate-script, .gemini/skills/generate-script, .github/skills/generate-script and .opencode/skills/generate-script in your project.

What does Generate Script need to run?

Going by SKILL.md and its folder, Generate Script needs the command-line tools its instructions call (python). Our summary lists: Python 3.

Does Generate Script access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Generate Script safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Generate Script use?

Generate Script is published under the AGPL-3.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Generate Script use?

About 1.4k tokens (SKILL.md is roughly 5.5k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Generate Script?

Skills that share tags, products or a category with Generate Script: Video Generation (bytedance/deer-flow, 83k stars), Video Shots (eternityspring/reelbench-skills, 868 stars), Video Cover Image (itwanger/toBeBetterJavaer, 18k stars) and Seedance (songguoxs/seedance-prompt-skill, 2.9k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Generate Script?

ArcReel (a GitHub organization) maintains it in ArcReel/ArcReel, which has 5,352 GitHub stars. The repository holds 18 skills in this directory. The repository was last updated on October 8, 2026.

Source: ArcReel/ArcReel on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.