APIPod MCP
Utilisez APIPod pour générer des images et des vidéos IA dans Claude, Claude Code, ChatGPT, Codex, Cursor, Grok et OpenClaw avec un seul compte APIPod — sans clés API de fournisseur.
Serveur MCP hébergé
npx add-mcp 'https://api.apipod.ai/mcp'S’installe dans Claude Code, Codex, Cursor et plus
Documentation
MCP · CLI—for agents and terminals
Generate AI images and video in Claude
Connect the remote MCP server to Claude, ChatGPT, Cursor or Codex, or install the CLI for terminals, scripts and agents that only run shell commands. Same models, same account, same USD prices.
MCPhttps://api.apipod.ai/mcp
any MCP client — no SDK, no proxy
CLInpm install -g apipod
terminals, scripts, CI — npm, Homebrew or pipx
OAuth 2.1 · PKCE S256 · 7 MCP tools · 1 CLI· plugin source on GitHub (Apache-2.0)· CLI source on GitHub (MIT)
claude.ai · 7 toolsPOST /mcp
you
Studio shot of this mug on a grey backdrop, 1:1 at 2K — and a 5-second clip panning across the same surface.
call
estimate_cost(2 jobs)
→ $0.05 + $0.40 = $0.45 · balance $18.20 · sufficient
call
generate_image(1:1, 2K) → img_2c91e4
generate_video(5s, 1080p) → vid_7f3a9c
done


0:05
Both jobs completed. Image 1:1 2K · video 5s 1080p.
completed · $0.45 · s.apipod.ai/…
2 jobs·check_job polled·failed jobs refunded
MCP server
For chat and IDE hosts
Claude, ChatGPT, Cursor, Codex and other MCP clients call seven tools by URL. Nothing to install.
- runs in
- the host, by URL
- sign-in
- OAuth in the host, once
- media in
- public HTTPS URLs
- results
- permanent URLs in the chat
CLI
For terminals, scripts and CI
One binary for macOS, Linux and Windows. Same commands for people and programs: tables by default, --json and exit codes for scripts.
- runs in
- any shell
- sign-in
- browser OAuth, or APIPOD_API_KEY
- media in
- local files or URLs
- results
- files on disk, or JSON
ClaudeconnectorWeb · DesktopClaude CodepluginCLIChatGPTconnectorWeb · DesktopCodexremote MCPCLI · IDECursormcp.jsonIDEGemini CLICLICLIGrokconnectorWeb · BotMuse Codesettings.jsonCLIOpenClawCLICLIWorkBuddymcp.jsonDesktopDeepSeek Harnessnative pluginWeb · CLITerminalAPIPod CLImacOS · Linux · Windows
Anything else with an MCP client and OAuth support uses the same URL.
Claude
hosted redirect
Hosted redirect to claude.ai/api/mcp/auth_callback. Nothing to install.
-
01
Settings → Connectors → Add custom connector.
-
02
Paste the URL and continue; Claude reads the auth metadata on its own.
-
03
Sign in with your APIPod account.
url Copy
https://api.apipod.ai/mcp
- redirect
- claude.ai/api/mcp/auth_callback
- endpoint
- api.apipod.ai/mcp
- auth
- OAuth 2.1 · PKCE S256
- scopes
- profile · email
npmHomebrewpipx
Copy
$ npm install -g apipod
$ apipod auth login
Installs the binary for your platform through optionalDependencies. No install script runs, so --ignore-scripts works.
Sign in once
- apipod auth login opens the browser and listens on 127.0.0.1 for one callback. Approve the APIPod CLI and the tokens go to the system keychain.
- For CI, set APIPOD_API_KEY. There is no --api-key flag: arguments end up in shell history and in ps.
- Credentials are bound to the host that issued them and never sent to another one. apipod auth logout revokes them on the server.
Built for scripts
--json prints exactly one document on stdout; progress goes to stderr. --no-wait returns the job id, and job wait picks it up later.
job=$(apipod video "a sunrise over the fjord" \
-m <model> --no-wait --max-cost 2 --json | jq -r .job_id)
apipod job wait "$job" -o ./out --json \
| jq -r '.results[].path'
Exit codes to branch on
- 0
- success
- 3
- not signed in, or the credential was rejected
- 6
- insufficient balance, over --max-cost, or spending not confirmed
- 7
- the wait timed out; the job keeps running
- 8
- the request or a local file was rejected
Full table, including 1, 2, 4 and 5, in the CLI README.
| capability | MCP | CLI | cost |
|---|---|---|---|
| Browse the catalog | list_models | apipod models list | free |
| Parameters a model accepts | get_model_params | apipod models params | free |
| Price quote vs. balance | estimate_cost | apipod cost | free |
| Remaining balance | get_balance | apipod balance | free |
| Generate or edit an image | generate_image | apipod image | paid |
| Generate a video | generate_video | apipod video | paid |
| Wait for a job, get results | check_job | apipod job wait | free |
| Upload a local file, get a URL | — (URLs only) | apipod upload | free |
Generation returns a job_id; check_job (MCP) or apipod job wait (CLI) returns permanent URLs and the final cost in USD. The CLI also downloads them.
Any other MCP client
If it speaks Streamable HTTP and OAuth, it works — add the server by URL:
{
"mcpServers": {
"apipod": {
"type": "http",
"url": "https://api.apipod.ai/mcp"
}
}
}
The server publishes its own auth metadata, so there is no client ID to paste.
Connect, verify, recover.
Three skills that answer the first ten minutes: what the server can do, one end-to-end run, and what to check when a call fails. - apipod - apipod-setup - apipod-troubleshootOne image, or one clip.
The request already names its outcome. These skills keep the agent from over-producing: no variant matrix for a single hero shot, no sequence for a five-second clip. - apipod-generate-image - apipod-generate-video - apipod-edit-image - apipod-vertical-videoSeveral outputs that have to agree.
A product set, an ad matrix, a creator-style ad, a multi-shot sequence. The hard part is continuity — the same object, the same person, the same light across every render. - apipod-product-shoot - apipod-ad-variations - apipod-ugc-ad - apipod-short-filmKnow the cost first.
Video is billed per second and the range between models is wide. These three skills produce a quote and a plan without spending anything. - apipod-estimate-cost - apipod-budget - apipod-plan-videoImage
Qwen Image 2.1$0.016 /call Qwen Image 3.0$0.03 /call Seedream 5.5$0.008 /call GPT Image 2.6$0.007 /call Wan 2.7 Image Spicy$0.03 /call Qwen Image Spicy$0.015 /call
+8 more in this category
Video
Seedance 2.5 Flash$0.041 /sec Seedance 2.5 Flash Spicy$0.042 /sec Gemini Omni Flash 1.2$0.023 /sec Gemini Omni Flash 1.2 Lite$0.02 /sec Gemini Omni Flash 1.1 Lite$0.30 /call Muse Video$0.15 /call
+30 more in this category
quote first
estimate_cost and apipod cost return a USD price for the exact request and whether the balance covers it.
a cap that holds
The CLI quotes before every generation and asks above $1. Without a terminal it refuses with exit code 6 unless --max-cost or --yes is given, so an agent cannot spend unattended.
refunded on failure
A provider-side failure returns the reservation to your balance automatically.
per-client grants
Each MCP client and the CLI hold their own OAuth grant. Revoke one under Authorized apps in the console without touching the rest.
Should I use the MCP server or the CLI?+
Use MCP when the work happens in a chat or IDE host such as Claude, ChatGPT or Cursor. Use the CLI for terminals, scripts, CI and agents that can only run shell commands. Both use the same account, models and prices, and you can use both.
Which clients are supported?+
Claude, Claude Code, ChatGPT, Codex, Cursor, Gemini CLI, Grok, Muse Code, OpenClaw and WorkBuddy — each with the exact config on this page. Anything else that speaks Streamable HTTP and OAuth works with the same URL. DeepSeek Harness uses a native plugin instead of MCP.
Which account does the client charge?+
The APIPod account you sign in with during OAuth. Generations land in your request log with path=/mcp, like any REST call.
Do I need an API key?+
Not for MCP: authentication is OAuth 2.1 with PKCE and nothing is pasted into the client. The CLI also signs in with OAuth; an API key in APIPOD_API_KEY is only for CI and other machines without a browser. The DeepSeek Harness plugin is the exception: it has no OAuth and always uses an API key.
Which models are available?+
Every image and video model in the catalog — Seedance, Kling, Seedream, Nano Banana, GPT Image, Wan and the rest. list_models returns what your account can call right now.
Is speech supported?+
Speech models appear in list_models, but generate_speech is not shipped yet — its output has to be persisted the same way images and video are. Use the REST API for TTS today.
Can I use my own images as references?+
Yes, on models whose get_model_params lists image input: pass public HTTPS URLs in image_urls — the first is the primary (edit source or first video frame). Base64 is not accepted over MCP.
What if my image is only on disk?+
With the CLI, pass the path: --image ./photo.png uploads it first, and apipod upload returns a URL for anything else. Over MCP the server takes URLs only, so upload the file in the console or with the CLI and pass the URL.
How long does a generation take?+
Images typically land in 10–40 seconds, video in 1–5 minutes. The client polls check_job for status; a finished job returns permanent URLs, so the result does not expire.
What can a client do?+
An MCP client reads the catalog and balance and submits generations, with profile and email scopes. The CLI additionally holds apipod:api, the scope that calls the REST API, and only the first-party CLI can request it. Every grant is revocable on its own under Authorized apps.
Is there a rate limit?+
Rate limits match your plan's REST API limits. There is no subscription for the MCP connection itself.
Create an account, then connect a client or install the CLI.
Sign in once from the client or with apipod auth login. No proxy to run, and no key to paste unless you want one for CI.
Create account Read the API docs
{"@context":"https://schema.org","@type":"Organization","name":"APIPod","url":"https://www.apipod.ai","logo":"https://www.apipod.ai/logo.svg","description":"AI media generation API for developers: top image and video models plus mainstream LLMs on one key, with multi-channel routing and automatic failover.","sameAs":["https://x.com/apipod_ai","https://github.com/apipod-ai","https://discord.gg/MusATnSzGx","https://t.me/apipod"],"contactPoint":{"@type":"ContactPoint","contactType":"Customer Support","email":"api@apipod.ai"}}
{"@context":"https://schema.org","@type":"WebSite","name":"APIPod","url":"https://www.apipod.ai","description":"The Ultimate AI Media API for Image & Video","potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://www.apipod.ai/models?search={search_term_string}"},"query-input":"required name=search_term_string"}}
{"@context":"https://schema.org","@type":"SoftwareApplication","name":"APIPod","applicationCategory":"DeveloperApplication","operatingSystem":"Web","description":"AI media generation API for developers: top image and video models plus mainstream LLMs on one key, with multi-channel routing and automatic failover."}
[{"@context":"https://schema.org","@type":"SoftwareApplication","name":"APIPod MCP Server","applicationCategory":"DeveloperApplication","operatingSystem":"Any","url":"https://www.apipod.ai/agent","description":"A remote MCP server over Streamable HTTP with seven tools for AI image and video generation: OAuth 2.1 with PKCE, USD pricing per generation.","featureList":["list_models","get_model_params","estimate_cost","get_balance","generate_image","generate_video","check_job"],"provider":{"@type":"Organization","name":"APIPod","url":"https://www.apipod.ai"}},{"@context":"https://schema.org","@type":"SoftwareApplication","name":"APIPod CLI","applicationCategory":"DeveloperApplication","operatingSystem":"macOS, Linux, Windows","url":"https://www.apipod.ai/agent#agent-cli","downloadUrl":"https://github.com/geniuspath/apipod-cli/releases","installUrl":"https://github.com/geniuspath/apipod-cli#install","softwareHelp":"https://github.com/geniuspath/apipod-cli","license":"https://opensource.org/licenses/MIT","offers":{"@type":"Offer","price":"0","priceCurrency":"USD"},"description":"A command-line client for AI image and video generation: one Go binary for macOS, Linux and Windows, JSON output, exit codes and spending caps for scripts and agents.","featureList":["apipod models list","apipod models params","apipod cost","apipod balance","apipod image","apipod video","apipod job wait","apipod upload"],"provider":{"@type":"Organization","name":"APIPod","url":"https://www.apipod.ai"}},{"@context":"https://schema.org","@type":"HowTo","name":"Connect your client.","description":"Every client reads the same two discovery documents: RFC 9728 at /.well-known/oauth-protected-resource/mcp, RFC 8414 at /.well-known/oauth-authorization-server. Hosted clients redirect to an allowlisted domain; CLIs call back on localhost, so no port has to be registered.","totalTime":"PT2M","step":[{"@type":"HowToStep","position":1,"name":"Copy the endpoint","text":"Copy https://api.apipod.ai/mcp. Every client uses the same URL."},{"@type":"HowToStep","position":2,"name":"Add it to the client","text":"Paste the URL into the client's connector or MCP config; the client reads the RFC 9728 protected-resource metadata and the RFC 8414 authorization-server metadata on its own."},{"@type":"HowToStep","position":3,"name":"Sign in","text":"Complete OAuth 2.1 with PKCE in the browser. The client then lists all seven tools."}]}]
{"@context":"https://schema.org","@type":"FAQPage","mainEntity":[{"@type":"Question","name":"Should I use the MCP server or the CLI?","acceptedAnswer":{"@type":"Answer","text":"Use MCP when the work happens in a chat or IDE host such as Claude, ChatGPT or Cursor. Use the CLI for terminals, scripts, CI and agents that can only run shell commands. Both use the same account, models and prices, and you can use both."}},{"@type":"Question","name":"Which clients are supported?","acceptedAnswer":{"@type":"Answer","text":"Claude, Claude Code, ChatGPT, Codex, Cursor, Gemini CLI, Grok, Muse Code, OpenClaw and WorkBuddy — each with the exact config on this page. Anything else that speaks Streamable HTTP and OAuth works with the same URL. DeepSeek Harness uses a native plugin instead of MCP."}},{"@type":"Question","name":"Which account does the client charge?","acceptedAnswer":{"@type":"Answer","text":"The APIPod account you sign in with during OAuth. Generations land in your request log with path=/mcp, like any REST call."}},{"@type":"Question","name":"Do I need an API key?","acceptedAnswer":{"@type":"Answer","text":"Not for MCP: authentication is OAuth 2.1 with PKCE and nothing is pasted into the client. The CLI also signs in with OAuth; an API key in APIPOD_API_KEY is only for CI and other machines without a browser. The DeepSeek Harness plugin is the exception: it has no OAuth and always uses an API key."}},{"@type":"Question","name":"Which models are available?","acceptedAnswer":{"@type":"Answer","text":"Every image and video model in the catalog — Seedance, Kling, Seedream, Nano Banana, GPT Image, Wan and the rest. list_models returns what your account can call right now."}},{"@type":"Question","name":"Is speech supported?","acceptedAnswer":{"@type":"Answer","text":"Speech models appear in list_models, but generate_speech is not shipped yet — its output has to be persisted the same way images and video are. Use the REST API for TTS today."}},{"@type":"Question","name":"Can I use my own images as references?","acceptedAnswer":{"@type":"Answer","text":"Yes, on models whose get_model_params lists image input: pass public HTTPS URLs in image_urls — the first is the primary (edit source or first video frame). Base64 is not accepted over MCP."}},{"@type":"Question","name":"What if my image is only on disk?","acceptedAnswer":{"@type":"Answer","text":"With the CLI, pass the path: --image ./photo.png uploads it first, and apipod upload returns a URL for anything else. Over MCP the server takes URLs only, so upload the file in the console or with the CLI and pass the URL."}},{"@type":"Question","name":"How long does a generation take?","acceptedAnswer":{"@type":"Answer","text":"Images typically land in 10–40 seconds, video in 1–5 minutes. The client polls check_job for status; a finished job returns permanent URLs, so the result does not expire."}},{"@type":"Question","name":"What can a client do?","acceptedAnswer":{"@type":"Answer","text":"An MCP client reads the catalog and balance and submits generations, with profile and email scopes. The CLI additionally holds apipod:api, the scope that calls the REST API, and only the first-party CLI can request it. Every grant is revocable on its own under Authorized apps."}},{"@type":"Question","name":"Is there a rate limit?","acceptedAnswer":{"@type":"Answer","text":"Rate limits match your plan's REST API limits. There is no subscription for the MCP connection itself."}}]}
{"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"APIPod","item":"https://www.apipod.ai"},{"@type":"ListItem","position":2,"name":"MCP & CLI","item":"https://www.apipod.ai/agent"}]}