DramaClaw - Industrialized Short-Drama Production Line
An end-to-end AIGC video engine that turns manuscripts into finished films — putting big-studio production power into the hands of individual creators.
- Source repo
- dramaclaw/dramaclaw
- Stars
- ★ 6.3k
- Last updated
- 3d ago
- License
- NOASSERTION
- Primary language
- Python
- FA score
- 0/100 · Major gaps
At a glance
- How it runs
- Works with
- Portable with changes
- Cost
- Free software; you pay for model usage
- Setup effort
- Medium · a few setup steps
- You'll need
- Typical use
- A web novelist who wants to turn a long-running series into multi-episode short dramas without manual script/storyboard work.
- Not a fit if
- Users unwilling to use a DC key or bring their own OpenAI-compatible endpoint
- Teams planning to resell it as a hosted service: Elastic 2.0 forbids this
- Source review
- 0/100 · Major gaps 8 safety controls not found
What does this agent do, and when should you use it?
DramaClaw is a source-available, industrialized short-drama production pipeline. Feed it a novel manuscript and it automates character extraction, episode planning, script generation, storyboard and first-frame art, voice-over synthesis, and final video editing. Built for creators, indie studios, and creative engineers, it can be self-hosted to run the entire drama factory on your own infrastructure — no stitching together a dozen tools or sending material to a black-box cloud. Core capabilities include novel parsing and story graphs, an asset library with identity consistency, episode and beat planning, multi-mode script generation, a line-art storyboard system, emotion-aware TTS, and video compositing and export. All inference goes through a remote OpenAI-compatible gateway, so the local footprint is light and no GPU is required for the standard pipeline.
DramaClaw parses a user-supplied novel into a queryable story graph (characters, relationships, timeline). Through its REST API (port 8780) and web UI (port 8080) it runs a series of async tasks: episode planning, script generation in adaptive/literal/staged modes, beat-driven storyboard and first-frame image generation, voice-over synthesis via IndexTTS2, and final assembly/export of episodes with subtitle (SRT) files and the full asset pack. It maintains an Asset Library for characters, scenes, props, and voices to lock identity across episodes, uses a Director World (3GS) to preserve spatial/camera consistency, and offers a Xia Director conversational assistant plus a Task Center for status, logs, cancel/retry, and resume-from-checkpoint. All state and media live on the local filesystem under the 'ce-data' volume — no Postgres, Redis, Celery, or Ray required.
- A web novelist who wants to turn a long-running series into multi-episode short dramas without manual script/storyboard work.
- A small ad agency that needs quick-turnaround product videos and short promo spots using the same pipeline with custom style templates.
- An otome game studio that uses the storyboard, TTS, and asset library to generate branching dialogue and scene assets.
- A creative engineer who wants to self-host the entire pipeline and plug in custom OpenAI-compatible models via a BYO gateway.
- A video production house that needs to keep character and scene consistency across many episodes, using Director World and the infinite canvas for iterative refinement.
How do you install or deploy this agent?
Recommended Docker approach: clone the repo, copy .env.example to .env and set PROMPT_EXPORT_PASSWORD, then run docker compose up -d --build (starts api and web services). For a no-build install, download the release compose file: curl -LO https://raw.githubusercontent.com/dramaclaw/dramaclaw/main/docker-compose.release.yml and run docker compose -f docker-compose.release.yml up -d. For local development you'll need Python 3.11–3.12, uv, and ffmpeg; then run uv sync and uv run novelvideo api --port 8780.
How do you use this agent?
Once running, open http://localhost:8080, go to Settings → Model Config → Official, and paste a DC key obtained from https://relayclaw.cdnfg.com (or configure your own OpenAI-compatible endpoint). Create a project, upload a manuscript, then step through the pipeline stages in order: parse, plan, script, storyboard, voice, composite. Each stage has its own UI and can be skipped or resumed. You can also drive everything programmatically through the REST API on port 8780 or ask the Xia Director agent to advance tasks conversationally.
What are this agent's strengths and limitations?
- End-to-end industrialized pipeline from script to finished film, with each step independently runnable and resumable.
- Self-hostable via Docker; media and state stored locally on filesystem for data control.
- Model-neutral: connects to any OpenAI-compatible text/image/video/audio gateway, with BYO option.
- Packages advanced features like identity/scene consistency (3GS), infinite canvas dual-track workflow, and the Xia Director agent.
- Default setup depends on DramaClaw's official gateway and a DC key, which may introduce cost or external dependency.
- Self-hosting a fully local model setup requires additional configuration and GPU/CUDA support only documented for optional features.
- The Elastic 2.0 license restricts reselling hosted service, which may limit some commercial use cases.
- Some advanced features (e.g., world model) are still evolving and may not be production-ready for all users.
How does this agent compare with similar options?
Compared to other tools in the space (e.g., L*TV, R*Hub, S*ko), DramaClaw differentiates with a full closed-loop short-drama production chain, including asset library, scene consistency, director world, Xia director agent, infinite canvas, and dual-track workflow, while being source-available and self-hostable.
Key facts side by side with the most closely related agents.
| Agent | Source review | Form / cost | Stars | Updated | Language | Full support on |
|---|---|---|---|---|---|---|
| DramaClaw - Industrialized Short-Drama Production Line This agent | 0 · Major gaps | Self-hosted serviceFree + model costs | ★ 6.3k | 3d ago | Python | — |
| Auto Browser | 62 · Some gaps | Self-hosted serviceFree + model costs | ★ 908 | today | Python | OpenAI API · Claude API |
| ByteChef | 53 · Major gaps | Self-hosted serviceFree + model costs | ★ 1k | today | Java | OpenAI API · Claude API |
| SandBase Harness | 52 · Major gaps | CLIFree + model costs | ★ 639 | 7d ago | TypeScript | OpenAI API · Claude API |
How does FollowAgents rate this agent?
Why each dimension lost points
Insufficient evidence: The repository does not provide explicit documentation on permission model, user confirmation mechanisms, data flow transparency, sensitive data handling, dependency security audits, external effect controls, rollback mechanisms, or source attribution. All trust criteria are scored 0 due to lack of evidence.
Insufficient evidence: No explicit evidence of self-consistency, dependency availability, or failure messages. Although test files exist, no execution results or reliability guarantees are provided.
Insufficient evidence: README describes broad scenarios but lacks detailed capability boundaries, trigger precision, or environment fit documentation.
Insufficient evidence: README provides installation instructions and documentation links, but lacks explicit evidence of information architecture, naming stability, examples/FAQ, known limitations, versioning/changelog, or maintenance responsibility. License metadata is NOASSERTION, but README declares Elastic License 2.0.
Insufficient evidence: README claims output usability and marginal value, but lacks cost-benefit analysis or actual output verification.
Insufficient evidence: Claims in README lack traceability, no cross-source corroboration, and no separation of facts and inferences.
- Not found in source: least-privilege scopingGrant only what the task needs: a dedicated account or read-only token, scoped to specific directories and repos.
- Not found in source: confirmation before actingTurn on (or add) a confirmation step before it acts, and try it in a sandbox or test environment before real data.
- Not found in source: data-flow disclosureWatch which external services it contacts (proxy or firewall logs) and keep sensitive data out until you know where it goes.
- Not found in source: sensitive-data handlingUse dedicated, low-privilege, revocable API keys — never production credentials — and keep secrets out of logs.
- Not found in source: dependency securityPin versions and run a dependency audit (npm audit, pip-audit) before installing; prefer running it in a container.
- Not found in source: disclosed external effectsEstablish which external systems it writes to, sends to or changes, and verify with test accounts or repos before production.
- Not found in source: rollback or recovery pathBack up first, or work on a git branch or snapshot, so its changes can be undone.
- Not found in source: verifiable attributionInstall from the official repo or registry and check the publisher and URL to avoid look-alike packages.
- The repository does not provide an AGENTS.md manifest, and the publisher identity is unverified; proceed with caution.
- License metadata is NOASSERTION, inconsistent with the Elastic License 2.0 declared in README; verify.
- Dependencies include many unpinned packages, posing supply chain risks.
- Functional claims in README lack verifiable test or execution evidence.
Recent releases
Latest entries from the project's GitHub Releases. Notes are quoted from upstream as written, not rewritten or tested by FollowAgents.
-
Release notes excerpt
v2.0.6 User-facing Highlights (zh) - 视频延长: 支持在已配置相应模型时延长画布中的视频，并保留原视频画幅。 - 视频画质与帧率增强: 可通过已配置的模型调整视频分辨率和帧率，并选择智能插帧等处理方式。 - 音视频分离输出 MP3: 提取的音频改为 MP3 格式，便于用作支持该格式的模型参考音频。 User-facing Highlights (en) - Video extension: Extend a canvas video when a compatible model is configured, while preserving its aspect ratio. - Video quality and frame-rate enhancement: Use configured models to adjust resolution and frame rate, including frame interpolation. …View on GitHub ↗ -
Release notes excerpt
v2.0.5 User-facing Highlights (zh) - 素材归档更完整: 视频、图片和音频生成结果支持归档后的交付与访问，减少临时结果链接失效带来的影响。 - 参考素材提前校验: 虾画生成前会检查参考素材的时长等限制，并标出不符合要求的文件，方便及时调整。 - 新增灵山图片模型: 社区版新增 LingShan G25 Fast 和 LingShan G25 Pro 图片模型配置。 - 账号与充值体验改进: 密码账号支持首次绑定手机号，并新增积分充值和独立结账流程。 User-facing Highlights (en) - More reliable archived media: Generated videos, images, and audio can be delivered from archives, reducing reliance on expiring result links. - Earlier reference validation: XiaHua checks reference media limits before generation and identifies files that need adjustment. …View on GitHub ↗ -
Release notes excerpt
--- attention: medium --- v2.0.4 User-facing Highlights (zh) - 新增越南语界面: DramaClaw 现已提供完整的越南语界面，并修正英文剧本格式指引和多语言回退逻辑，让更多社区用户可以直接使用熟悉的语言完成创作。 - 支持可选手机号验证码登录: 部署方启用对应入口后，用户可以通过手机号和验证码登录并设置密码；登录能力默认受开关控制，不影响现有部署。 - 跨项目媒体引用更安全: 虾画会识别来自其他项目的图片和视频引用，阻止无效引用继续写入，并为历史遗留引用提供复制到当前项目的一键修复入口。 - 本地 CE 启动更省配置: 本地启动现在默认使用社区版，不再要求额外声明版本；并发初始化配置时也会自动重试。 - 画布任务计费参数更加准确: 画布主线生成任务会携带完整的模型计费参数，减少预估与实际结算不一致的情况。 User-facing Highlights (en) …View on GitHub ↗