Integrations

Codex SDK

Set up GenAsset for Codex, test workspace connectivity, and run a generate-then-save flow.

Setup

From zero, use this Codex SDK terminal flow:

Codex SDK setup

bash

# 1) Check Codex CLI is installed
codex --version

# 2) Add GenAsset MCP server
codex mcp add genasset --url https://genasset.xyz/api/mcp

# 3) Confirm it is registered
codex mcp list

# 4) Optional: inspect this server config
codex mcp get genasset

Optional bearer-token setup

Use this if you want calls like workspace_status without passing workspace_token each time.

Codex SDK with bearer token

bash

# Optional: avoid passing workspace_token in every prompt
export GENASSET_WORKSPACE_TOKEN="ga_your_workspace_token"

# Re-add MCP server with bearer token support
codex mcp remove genasset
codex mcp add genasset \
  --url https://genasset.xyz/api/mcp \
  --bearer-token-env-var GENASSET_WORKSPACE_TOKEN

codex mcp list

Test connection prompt

After restarting Codex, run this test prompt.

First test prompt

text

Call workspace_status with {"workspace_token":"ga_your_workspace_token"}.
Return workspace name and plan.

If you configured bearer token env, you can call:
Call workspace_status with {}.

End-to-end prompt

Run one full generate-then-save flow and verify by loading the newest recipe.

End-to-end Codex prompt

text

Generate an image of a cinematic woman in rain.
Then call save_image_to_genasset with:
- asset_name: "mcp-codex-test-1"
- workspace_token: "ga_your_workspace_token" (optional if env token is set)
- model: "gpt-image-1"
- source: "codex"
- prompt: "cinematic woman in rain, clean portrait, blue coat"

Then call list_assets search "mcp-codex-test-1" and load_version_recipe for the newest version.

Notes

Codex Desktop note (macOS)

If you use the GUI app, set the token in launchd so the app process can read it:

Set desktop env var

bash

launchctl setenv GENASSET_WORKSPACE_TOKEN "ga_your_workspace_token"

To remove it later:

Unset desktop env var

bash

launchctl unsetenv GENASSET_WORKSPACE_TOKEN