Axiom Lab — AI Pitch Deck & Business Video MCP

io.github.Axiom-lab-tech由 GitHub 帳號 Axiom-lab-tech 發布v0.6.0更新於 Oct 25, 2026

Turn PDFs, PPTX decks, product photos, and briefs into QA-checked commercial videos via MCP.

發布方自己的服務：Axiom

已驗證Streamable HTTP可網頁執行效率與工作流媒體與設計

服務詳細資訊

識別碼: io.github.Axiom-lab-tech/axiomlab-mcp
傳輸: Streamable HTTP
執行環境: Web executable
信任: Verified
發布於: Oct 25, 2026

安全提示

MCP 服務會接收對話中的工具呼叫請求，並可能執行外部操作。啟用前請先了解該服務能做什麼，並只授予它所需的權限。

註冊中心已驗證此條目所屬的 GitHub 帳號 Axiom-lab-tech。這並不表明該帳號就是產品方。

SourceWeft 的檢查只涵蓋條目是否完整、端點是否回應；不審查伺服器的程式碼，也不審查其工具的行為。

在 README 中新增徽章

向讀者展示該 MCP 服務已收錄於 SourceWeft，徽章連結到本頁。

[![Listed on SourceWeft](https://sourceweft.com/badge/mcp/io.github.Axiom-lab-tech%2Faxiomlab-mcp.svg)](https://sourceweft.com/mcp/io.github.Axiom-lab-tech%2Faxiomlab-mcp?utm_source=readme&utm_medium=badge)

概覽

AI 產生的概覽

遠端 MCP 伺服器，可將簡報、投影片、產品照片與需求文件轉成經品管檢查的商業影片，付費算圖前先免費產生方案。

可以做什麼: 列出影片工作流程、範本與影片類型，並瀏覽現成的範例與相關資源。上傳 PDF、PPTX、DOCX 需求文件、圖片、影片片段或音訊，並列出已上傳素材及其 ID。在付費算圖前依素材產生免費的結構化製作方案。啟動、查詢、長輪詢、列出、取消、恢復影片任務或確認長度，並在任務要求時核准額外預算。查詢預付錢包餘額。
適用情境: 適合想讓助理把既有行銷素材（例如簡報、產品照片或文字需求文件）製作成短影片或解說影片，並希望在付費算圖前先檢視免費方案的情境。
執行需求: 需要遠端 Streamable HTTP 端點，以及在服務商開發者頁面建立的 bearer API 金鑰，透過 Authorization 標頭送出。README 另說明可選的本機 stdio 伺服器，需以 pip 或 uvx 安裝，依賴 Python，並以環境變數提供同一組 API 金鑰。

安裝前請注意

影片製作需付費，從預付的港幣錢包扣款；可透過預算核准工具授權額外支出。上傳的簡報、照片、影片片段與音訊會傳送到服務商的服務。API 金鑰屬於機密，應避免寫入共用設定；viewer 角色的金鑰只能檢視，不能消費。

安裝

在 SourceWeft 中

開啟儀表板中的 Axiom Lab — AI Pitch Deck & Business Video MCP，將其新增到工作區。
連接 Authorization token 憑證。它們會被私密儲存，絕不會在公開頁面上展示。
為需要使用其工具的對話啟用該服務。

Web executable，透過 Streamable HTTP。遠端服務在工作區中設定後即可從網頁執行環境執行。

交給代理程式安裝

把這段提示詞貼上到 Claude Code、Codex、Cursor 等代理程式，或 SourceWeft 對話中。代理程式會讀取該伺服器的安裝指南，先向你展示發布者、端點和登入方式，經你同意後再把它新增到你的客戶端。金鑰不會寫進任何命令。

閱讀 https://sourceweft.com/mcp/io.github.Axiom-lab-tech%2Faxiomlab-mcp/install.md，並按其中說明把這個 MCP 伺服器接入我的客戶端。修改任何設定之前，先告訴我它的發布者、端點和登入方式，等我確認。不要把 API 金鑰寫進命令或我會提交的檔案；問我金鑰應從哪裡取得。

檢視代理程式所依據的安裝指南

其他 MCP 客戶端

把它新增到你客戶端的 mcpServers 設定中。

{
  "mcpServers": {
    "axiom-lab-ai-pitch-deck-business-video": {
      "type": "http",
      "url": "https://axiomlab.online/mcp",
      "headers": {
        "Authorization": "Bearer <YOUR_TOKEN>"
      }
    }
  }
}

README

Axiom Lab MCP Server, CLI & SDKs (`azimo`)

[Axiom-lab-tech/axiomlab-mcp MCP server]

Official Model Context Protocol (MCP) server, azimo command-line client, Python SDK, and TypeScript SDK for Axiom Lab — turn PDF pitch decks, PPTX slides, UI screenshots, product photos, and written briefs into QA-checked commercial videos.

Website: https://axiomlab.online
Official MCP Registry: io.github.Axiom-lab-tech/axiomlab-mcp
Glama MCP Directory: https://glama.ai/mcp/servers/Axiom-lab-tech/axiomlab-mcp
Smithery Skill: axiomlab/azimo
MCP & Agent Guide: English · 中文文档
REST API Reference: https://axiomlab.online/developers/api
Claude Skill (SKILL.md): skills/azimo/SKILL.md · Online Guide
Full Playable Examples: https://axiomlab.online/examples
Free Pitch & Startup Calculators: https://axiomlab.online/tools

Two Ways to Connect via MCP

Option 1: Remote Streamable HTTP MCP (Zero Local Install)

Point Claude Code, Cursor, or any Streamable HTTP MCP client directly at https://axiomlab.online/mcp using an API key from https://axiomlab.online/dev:

bash

claude mcp add --transport http azimo https://axiomlab.online/mcp \  --header "Authorization: Bearer $AZIMO_API_KEY"

claude mcp add --transport http azimo https://axiomlab.online/mcp \  --header "Authorization: Bearer $AZIMO_API_KEY"

Or in ~/.cursor/mcp.json / claude_desktop_config.json:

json

{  "mcpServers": {    "azimo": {      "url": "https://axiomlab.online/mcp",      "headers": {        "Authorization": "Bearer ${env:AZIMO_API_KEY}"      }    }  }}

{  "mcpServers": {    "azimo": {      "url": "https://axiomlab.online/mcp",      "headers": {        "Authorization": "Bearer ${env:AZIMO_API_KEY}"      }    }  }}

Option 2: Local `stdio` MCP Server (`azimo mcp`)

Run the local stdio MCP server directly on macOS, Linux, or Windows (enables direct local file path uploads up to 60 MB per file without base64 inline limits):

Configure in claude_desktop_config.json or ~/.cursor/mcp.json:

json

{  "mcpServers": {    "azimo": {      "command": "azimo",      "args": ["mcp"],      "env": {        "AZIMO_API_KEY": "vg_..."      }    }  }}

{  "mcpServers": {    "azimo": {      "command": "azimo",      "args": ["mcp"],      "env": {        "AZIMO_API_KEY": "vg_..."      }    }  }}

Or launch ephemeral via uvx:

Available MCP Tools (14 Tools, 4 Prompts, 2 Resources)

Tool	Description
`azimo_list_workflows`	List supported video workflows (`explainer`, `motion_explainer`, `product_pitch`, `profile_card`, `short_drama`, `creative_video`), durations, and aspect ratios
`azimo_list_playground`	List ready-made starting templates (project intro, person card, product film, company intro, short-drama ad, e-commerce ad) with runnable examples
`azimo_list_video_types`	Inspect structured fields for `project`, `product`, `personal`, `corporate`, and `name_card` videos
`azimo_upload_asset`	Upload source PDFs, PPTX decks, DOCX briefs, PNG/JPG/WEBP photos, MP4/MOV clips, or audio references
`azimo_list_assets`	List files uploaded to the current project with their `asset_id`s
`azimo_create_plan`	Validate material and generate a free structured production plan before any paid rendering starts
`azimo_create_video`	Start paid video production from a confirmed `ready` plan (idempotent per `plan_id`)
`azimo_get_video`	Retrieve job status, stage, progress, and signed MP4 film + cover (+ SRT) download links
`azimo_wait_video`	Long-poll until production reaches a terminal state or pauses at `needs_input` (`confirm_duration`)
`azimo_list_videos`	List recent videos in the active project with status
`azimo_cancel_video`	Cancel a queued, waiting, or running video job before its next paid step
`azimo_approve_budget`	Authorize extra spending when a video stops with an approvable `budget_approval` request
`azimo_resume_video`	Confirm the suggested film duration (`duration_sec`, 30–180s) or continue a paused `needs_input` video
`azimo_get_balance`	Check total, reserved, and available prepaid HKD wallet balance

MCP Prompts (4): create_product_video, create_company_intro, create_project_intro, create_person_card
MCP Resources (2): azimo://workflows, azimo://video-types

Safety by design: Plans are always free. Production and billing only start after explicit confirmation, and viewer-role API keys can inspect plans and progress without permission to spend balance.

Python SDK

python

import osfrom vidgen_sdk import Vidgen
with Vidgen("https://axiomlab.online", os.environ["AZIMO_API_KEY"]) as api:    asset = api.upload_asset("deck.pdf")    plan = api.create_plan(        {            "prompt": "Explain in 60 seconds how our inventory platform cuts warehouse stockouts",            "assets": [asset["id"]],            "options": {"workflow": "auto", "duration_max": 60},        },        idempotency_key="launch-plan-001",    )    if plan["data"]["status"] == "ready":        job = api.create_video(plan_id=plan["id"], idempotency_key="launch-video-001")        final = api.wait(job["id"])        stop = (final.get("result") or {}).get("input_required") or {}        if stop.get("code") == "confirm_duration":            api.confirm_duration(job["id"], stop["suggested_duration_sec"], idempotency_key="launch-length-001")            final = api.wait(job["id"])

import osfrom vidgen_sdk import Vidgen
with Vidgen("https://axiomlab.online", os.environ["AZIMO_API_KEY"]) as api:    asset = api.upload_asset("deck.pdf")    plan = api.create_plan(        {            "prompt": "Explain in 60 seconds how our inventory platform cuts warehouse stockouts",            "assets": [asset["id"]],            "options": {"workflow": "auto", "duration_max": 60},        },        idempotency_key="launch-plan-001",    )    if plan["data"]["status"] == "ready":        job = api.create_video(plan_id=plan["id"], idempotency_key="launch-video-001")        final = api.wait(job["id"])        stop = (final.get("result") or {}).get("input_required") or {}        if stop.get("code") == "confirm_duration":            api.confirm_duration(job["id"], stop["suggested_duration_sec"], idempotency_key="launch-length-001")            final = api.wait(job["id"])