Skip to main content
Connect a host that supports Streamable HTTP to:
No Node.js or npm process is needed for remote access. The server is stateless at the connection level; paid jobs and recovery receipts remain on the backend. Server changes require a backend deployment, and your host may need to reconnect or refresh its cached tool list. Remote does not mean offline or deployment-free.

​
Compose inside another workflow

Keep the upstream script, prompt, model/provider, count and approved budget. Creative helpers are optional. Use persistent UUID requestId, wait: false, structured results and check_generation to submit frames/videos and resume them without re-approving every step. The caller owns bounded scheduling, previews and downloads. See the complete workflow contract and examples.

​
Account and authentication

Create a MeiGen key using a desktop browser at API Keys. In the same account, choose Top Up on Profile, or use the mobile premium page. Configure the secret in the host’s private credentials/headers settings:
Paid calls spend purchased credits only; no daily free credits or free Skill attempts. Do not place the key in chat or in the URL.

​
Codex CLI, IDE extension and local Codex desktop host

Equivalent ~/.codex/config.toml entry, with Skill timeouts:
Merge into the existing entry rather than adding a duplicate. Set the variable in the environment that launches Codex; a project .env.local is not automatically loaded. To verify public lookups before setting a key, omit the token flag/TOML field. If a desktop launch cannot inherit the variable, use its private Authorization-header setting when supported, or privately configure http_headers.Authorization and remove bearer_token_env_var. Preserve other servers and keep secrets out of shared configuration. See the official Codex MCP guide.

​
Claude Code and other HTTP hosts

For Claude Code, use its private local terminal/connection settings:
Other hosts often use the JSON shape below; field names vary by client. Replace the placeholder privately in your host’s credentials settings.

​
ChatGPT web: public lookups only

For an account/workspace with custom MCP access, enable Developer mode, create a custom remote app/plugin for the URL above, and choose No Authentication. Follow OpenAI’s current setup. ChatGPT’s hosted client cannot send custom API keys; MeiGen has not provided the OAuth integration needed for authenticated use there. You can use public tools, including list_skills, list_models, gallery lookups and check_generation for a known generation ID. Generation, upload, requestId recovery and check_skill require a key and are unavailable through this ChatGPT web connection. Codex has a different configuration flow above. A chat instruction cannot install npm inside ChatGPT web.

​
Tools

The remote server exposes 14 tools:

​
First successful check

Reconnect, then ask: “List MeiGen’s available Skills and current prices.” Verify that an actual tool call succeeds; registration alone is not connection proof. It does not generate or spend credits. Then supply the required material and request one of the five workflows. Skills return after submission and use check_skill; generic generation recovery uses check_generation. Remote tools cannot read a local filesystem path. Use a public image URL or upload_skill_image with actual accessible bytes. For Upscale, pass the original URL directly or upload with purpose: "upscale"; do not downsize it as an ordinary reference first.

​
GPT Image 2.5 options

When list_models includes gpt-image-2.5, generate_image accepts all five quality levels and modelVariant: "sunburst" (quality priority, default) or "flare" (speed priority). The two variants have the same credit price. Example tool arguments:
Check the current model price before submitting this 2K Standard request. Add public reference-image URLs with referenceImages for editing or visual guidance. The UUID shown is illustrative: allocate and persist your own requestId once per logical attempt, then reuse it only with the exact inputs for recovery. wait: false returns after submission. Use check_generation with requestId or generationId after an interruption. Legacy attemptId remains accepted; new integrations should use requestId. See model pricing for the full matrix.

​
Remote and local

Existing keys and the endpoint remain usable across server updates. Switching from local to remote is optional; local-only capabilities do not migrate. See local setup and upgrade notes. upload_skill_image requires a positive purchased-credit balance and does not charge for generation. For Upscale attachments set purpose: "upscale". Every upscale_image MCP call, including the first, requires confirmedCredits from the current list_skills quote within the accepted user or upstream workflow budget; reuse explicit acceptance.