Skip to content
OpenAgentSkill
Browsing is open · Catalog captured 2026-11-21 · Automatic updates and account services are paused.
Development

narrator-ai-cli

Create AI-narrated film/drama commentary videos via CLI. Two workflow paths (Original & Adapted narration), 93 movies, 146 BGM tracks, 63 dubbing voices in 11 languages, 90+ narration templates. Use when creating narration videos, film commentary, short drama dubbing, or video production.

NarratorAI-Studio · Static checks on file

Installation

Read the source and dependencies before running a command. Copying does not install or execute anything.

npx skills add NarratorAI-Studio/narrator-ai-cli --skill narrator-ai-cli

About this skill


name: narrator-ai-cli version: "1.0.0" license: MIT description: >- Create AI-narrated film/drama commentary videos via CLI. Two workflow paths (Original & Adapted narration), 93 movies, 146 BGM tracks, 63 dubbing voices in 11 languages, 90+ narration templates. Use when creating narration videos, film commentary, short drama dubbing, or video production. user-invocable: true tags:

  • video-narration
  • film-commentary
  • ai-video
  • short-drama
  • content-creation
  • dubbing
  • tts
  • video-production metadata: openclaw: emoji: "🎬" requires: bins: - narrator-ai-cli primaryEnv: - NARRATOR_APP_KEY

narrator-ai-cli Skill

You have access to narrator-ai-cli, a CLI tool for the Narrator AI video narration API. Always use --json flag for all commands to get structured output you can parse.

Prerequisites

The CLI must be configured before use. Check with:

narrator-ai-cli config show

If not configured, set up with:

narrator-ai-cli config set app_key <app_key>

Server defaults to https://openapi.jieshuo.cn — no need to configure.

Need an app_key or help with setup? Contact us to get started:

  • 📧 Email: merlinyang@gridltd.com
  • 💬 WeChat:

Core Rules

  1. ALWAYS use --json for all commands. Parse the JSON output to extract values for next steps.
  2. NEVER fabricate data — never guess file IDs, movie info, or voice IDs. Always query from CLI commands.
  3. ASK user to choose when there are multiple options (pre-built vs custom, which template, which movie, etc.).
  4. Poll task status — tasks are async. After creation, poll with narrator-ai-cli task query <task_id> --json until status is 2 (success) or 3 (failed).
  5. video-composing order_num source differs by path:
    • Standard Path: use generate-writing's task_order_num (NOT clip-data's)
    • Fast Path: use fast-clip-data's task_order_num
  6. dubbing list returns id and type — pass id as dubbing, pass type as dubbing_type. They are two separate fields.
  7. negative_oss_key = video_oss_key — always set negative_oss_key to the same value as video_oss_key in episodes_data.
  8. search-movie may take 60+ seconds — this is normal (Gradio backend).

Decision Points (ASK USER)

Before starting any workflow, you need to help the user make the following choices. Present the options and let the user decide. Do NOT pick for them.

Decision 1: Which workflow path?

ASK: "请选择工作流路径:"

  • 二创文案: 爆款学习 → 生成解说文案 → 生成剪辑数据 → 合成视频 → 视觉模板(可选)
    • Based on existing reference video style, generates adapted narration
    • Suitable when: user wants to recreate content based on reference material
  • 原创文案 (faster & cheaper): 快速文案 → 快速剪辑数据 → 合成视频 → 视觉模板(可选)
    • Generates original narration from movie info + narration template
    • Three sub-modes: 热门影视(hot drama), 原声混剪(original mix), 冷门/新剧(new drama)
    • Suitable when: user wants original narration, faster speed, lower cost

Decision 2: Narration style — pre-built template or custom learning?

ASK: "解说风格:使用预置模板还是自定义学习?"

  • Option A: Pre-built template (recommended, 90+ templates across 12 genres)

    narrator-ai-cli task narration-styles --json
    # or filter: narrator-ai-cli task narration-styles --genre <genre> --json
    

    Show the list to user, let them pick one. The id field is the learning_model_id.

  • Option B: Custom learning (popular-learning task, Standard Path only) Requires user to provide a reference video + SRT. The output learning_model_id is used in generate-writing.

Decision 3: Movie source files — pre-built material or user upload?

ASK: "电影素材:使用预置素材还是自己上传的文件?"

  • Option A: Pre-built materials (93 movies ready to use)

    narrator-ai-cli material list --json
    # or search: narrator-ai-cli material list --search "<movie_name>" --json
    # or filter: narrator-ai-cli material list --genre <genre> --json
    

    Show results to user. Each entry has video_id and srt_id.

    • Use video_id as video_oss_key AND negative_oss_key
    • Use srt_id as srt_oss_key
  • Option B: User's uploaded files

    narrator-ai-cli file list --json
    # or search: narrator-ai-cli file list --search "<keyword>" --json
    # or transfer by link: narrator-ai-cli file transfer --link "<url>" --json
    

    User picks video and SRT files from their cloud storage.

Decision 4: BGM — pre-built track or user upload?

ASK: "背景音乐:使用预置BGM还是自己上传的音频?"

  • Option A: Pre-built BGM (146 tracks)

    narrator-ai-cli bgm list --json
    # or search: narrator-ai-cli bgm list --search "<keyword>" --json
    

    Show results to user. The id field is the bgm parameter value.

  • Option B: User's uploaded audio

    narrator-ai-cli file list --json
    

    User picks an audio file. Use its file_id as the bgm parameter.

Decision 5: Dubbing voice — which voice and language?

ASK: "配音角色:请选择配音语言和角色"

# First show available languages
narrator-ai-cli dubbing languages --json

# Then list voices for chosen language
narrator-ai-cli dubbing list --lang <language> --json

# Or filter by genre recommendation
narrator-ai-cli dubbing list --tag <genre_tag> --json

Show results to user. From the selected voice:

  • id field → pass as dubbing parameter
  • type field → pass as dubbing_type parameter

There is no "custom upload" option for dubbing — always pick from the pre-built list. (Users can create custom voices via voice-clone task separately.)

Decision 6: Visual template (optional)?

ASK: "是否需要应用视觉模板?(可选步骤)"

  • Yes: Pick a template
    narrator-ai-cli task templates --json
    
    Show results to user, let them pick. The name field goes into template_name list.
  • No: Skip magic-video step.

Workflow Path 1: 二创文案

popular-learning(optional) -> generate-writing -> clip-data -> video-composing -> magic-video(optional)
`
Read the complete source ↗