mcpField
D
dramapack

dramapack

Local
dawnwood.aiOther0Updated 1d ago
RepositoryVisit website

About dramapack

dramapack is a command-line and MCP server for preparing short-drama episodes for platform delivery. It combines mastering, subtitle handling, rights checks and local quality control in one delivery workflow.

Key features

  1. 01Platform-ready delivery packages
  2. 02-14 LUFS audio mastering
  3. 03Subtitle safe-area burning
  4. 04Clean masters and 3:4 covers
  5. 05Local QC and rights checks

Use cases

  1. 01Preparing episodes for platform submission
  2. 02Checking audio levels and delivery specs
  3. 03Finding black frames or subtitle drift
  4. 04Reviewing music and font rights

MCP support

As an MCP server, dramapack exposes inspect, master, package, audit and check_rights tools to AI agents. Results include JSON status fields.

Configuration

Local server config is the same for every client. Copy it into your client’s MCP config file.

mcp.json · Universal config
{
  "mcpServers": {
    "dawnwood-ai-dramapack": {
      "command": "uvx",
      "args": [
        "dramapack"
      ],
      "env": {
        "DRAMAPACK_FFMPEG": "<DRAMAPACK_FFMPEG>",
        "DRAMAPACK_FFPROBE": "<DRAMAPACK_FFPROBE>",
        "DRAMAPACK_VLM_BASE_URL": "<DRAMAPACK_VLM_BASE_URL>",
        "DRAMAPACK_VLM_API_KEY": "<DRAMAPACK_VLM_API_KEY>",
        "DRAMAPACK_VLM_MODEL": "<DRAMAPACK_VLM_MODEL>"
      }
    }
  }
}
Run commanduvx dramapack
Environment variables to set
DRAMAPACK_FFMPEGPath to ffmpeg (default: ffmpeg on PATH)
DRAMAPACK_FFPROBEPath to ffprobe (default: ffprobe on PATH)
DRAMAPACK_VLM_BASE_URLOptional: OpenAI-compatible base URL for audit vlm checks (your provider or a local model)
DRAMAPACK_VLM_API_KEYSecretOptional: API key for that endpoint (any non-empty string for local servers)
DRAMAPACK_VLM_MODELOptional: vision-capable model name for audit vlm checks

Local servers need a runtime on your machine (e.g. Node.js 18+). Config file locations differ by client; see your client’s docs.

Packages
pypidramapack0.2.1

Tools

No tool list yet

Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.

Details

From the project README

dramapack

tests License: Apache-2.0 MCP server

dramapack turns short-drama episodes (vertical dramas, micro dramas, AI dramas) into a platform-ready delivery package — TikTok Drama Center / DramaWave / ReelShort submission pack, -14 LUFS mastering, burned subtitles inside the subtitle safe area, clean masters, 3:4 covers, local QC (black frames, subtitle/speech alignment, optional bring-your-own vision-model checks) and music/font rights check — from the command line or as an MCP server for AI agents.

Read in Chinese · Docs · Reference

Why

Between "the episode is rendered" and "the platform accepted it", the same things go wrong for every short-drama studio:

  • Submission packs: numbered episodes, same-name SRTs, 3:4 covers, a clean master, edit evidence, chain of title, AI disclosure. Dozens of manual steps per series. → delivery package guide
  • Phone-speaker loudness: raw AI renders sit at -20 to -27 LUFS; the feed around them is near -14. And ffmpeg loudnorm silently goes dynamic on dialogue and lifts the noise floor by 30-50 dB. → -14 LUFS guide
  • Subtitle safe areas: ffmpeg's default 384x288 ASS canvas scales every size ~6.7x; fixed margins land under the player bar or exceed TikTok Drama Center's 70%-width rule. → safe-area guide
  • Clean masters that are really the same edit as the subtitled one. → clean master guide
  • Rights hygiene: no major-label music, commercial-free fonts only. → rights checklist
  • QC before upload: black frames mid-episode, subtitles that drift from the speech, speech nobody subtitled, likely subtitle typos; optionally AI visual defects (hands, faces, identity drift, garbled text) and content-policy flags with your own vision model or a local one. Free, local, no account; no network calls unless you opt in. Lip-sync is not covered. → local QC guide

Install

Requires Python >= 3.10 and ffmpeg (with libass + libx264) on PATH. For CJK subtitles install Noto CJK fonts (apt install fonts-noto-cjk).

pip install dramapack
# or run without installing:
uvx dramapack --help

Optional dialogue-density metrics and subtitle/speech alignment: pip install "dramapack[speech]".

60-second quickstart

# input/ holds 01.mp4 02.mp4 ... (un-subtitled cuts) and 01.srt 02.srt ...
dramapack check-rights --input input/          # music + fonts: fix red first
dramapack package input/ out/ --title "My Show" --studio "My Studio" --bed tension
dramapack audit out/episodes/01.mp4 --cut input/01.mp4   # QC: black, subtitles, pacing
dramapack inspect out/episodes/01.mp4          # LUFS, true peak, specs

out/ now holds episodes/ (subtitled, -14 LUFS) with same-name .srt, clean/ (no subtitles, same edit), covers/ (3:4), timeline/, docs/ (chain of title to sign, AI disclosure), manifest.json and a README for the recipient. Add --platform-pairs for two-episodes-per-file cuts. Optional show.json / rights.json / covers / timelines: see examples/input/ and the reference.

For AI agents

claude mcp add dramapack -- dramapack mcp

MCP tools: inspect, master, package, audit, check_rights. Each returns JSON with ok and a green/yellow/red status. Every CLI subcommand also takes --json; exit codes are stable (0 ok, 1 red, 2 usage, 3 failure). The repo doubles as a Claude Code plugin (.claude-plugin/, skills/dramapack/) and has an MCP Registry entry (server.json). → agent guide · AGENTS.md · llms.txt

Docs

Please credit

dramapack is released under the Apache License 2.0. Use it for anything, including commercial work. We'd appreciate two things. These are requests, not license conditions:

  • Keep the generator credit dramapack writes into delivery packs ("generator" in manifest.json and the one-line "Packaged with dramapack by Dawnwood" footers).
  • Mention "dramapack by Dawnwood" when you describe a production pipeline that uses it.

See NOTICE. To cite it: CITATION.cff.

Built and used in production by Dawnwood, an AI drama studio — https://dawnwood.ai/open-source/

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.