BeatAPI Codex Plugin
Enables creating and managing BeatAPI AI Music Video and Ecommerce Video workflows. Supports inspecting workflows, credits, usage, uploading files, creating tasks, polling tasks, and managing webhooks.
README
BeatAPI Codex Plugin
Create and manage BeatAPI AI Music Video and Ecommerce Video workflows directly
from Codex. The plugin combines the canonical beatapi-video Skill with a
bundled local MCP server and uses the same API key as the BeatAPI CLI.
What users can do
- inspect workflows, credits, usage, and concurrency;
- upload local images, audio, and SRT files;
- create automatic or manual Music Video tasks;
- inspect, edit, materialize, and compose storyboard shots;
- create Ecommerce Video tasks;
- poll asynchronous tasks until a terminal or actionable state;
- create, inspect, update, and delete webhook endpoints.
The plugin does not put API keys in prompts or MCP tool arguments. It first uses
BEATAPI_API_KEY; otherwise its local MCP server invokes the installed
beatapi CLI, which reads the key saved by beatapi auth login from the
operating-system credential manager.
Install for Codex desktop
Prerequisites:
- Node.js 20.19+ or 22.12+;
- Codex/ChatGPT desktop with plugin support;
- a BeatAPI account and API key from BeatAPI.
From a source checkout:
npm ci
npm run verify
codex plugin marketplace add ./dist/marketplace
codex plugin add beatapi-codex-plugin@beatapi-local
Then authenticate once in a terminal:
npm install --global beatapi
beatapi auth login
Alternatively, set BEATAPI_API_KEY in the environment that launches Codex.
Never paste the key into a conversation.
Restart the desktop app after installation. Useful starter requests include:
- “Use
$beatapi-videoto create a music video from my images and audio.” - “Turn these product photos into a 15-second 9:16 ad.”
- “Check my BeatAPI credits and the status of task
task_....”
Package layout
.codex-plugin/plugin.json— Codex presentation and component manifest..mcp.json— local stdio MCP configuration.mcp/server.mjs— dependency-free bundled MCP runtime.skills/beatapi-video/— synchronized canonical BeatAPI Skill.contract/— locked BeatAPI OpenAPI snapshot.generated/— Skill and client-runtime provenance locks.submission/— official Plugin Directory listing and review materials.
Do not edit skills/beatapi-video or mcp/vendor/client directly:
npm run skill:sync
npm run runtime:sync
Publishing paths
This repository supports two distinct release paths:
- Codex desktop/local marketplace.
npm run marketplace:buildcreates a complete installable marketplace and ZIP underdist/. - Public OpenAI Plugin Directory.
npm run submission:buildcreates a Skills-only ZIP that can be uploaded to the official submission portal. This artifact contains the Skill but not the local MCP server, so users need the globally installedbeatapiCLI unless their host supplies compatible BeatAPI MCP tools.
The local plugin includes a stdio MCP server. Official MCP-backed public review requires a separately deployed public HTTPS MCP server, domain verification, and reviewer authentication. This repository does not claim that hosted infrastructure; see submission/SUBMISSION.md.
Verification
npm run verify
python3 ~/.codex/skills/.system/plugin-creator/scripts/validate_plugin.py .
Verification checks OpenAPI drift, synchronized Skill/client sources, TypeScript, MCP protocol behavior, credential redaction, webhook-secret file permissions, deterministic bundles, marketplace packaging, and submission packaging.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。