| image-generation-settings |
Developer-only image generation settings and host API |
gateway-backend-communication |
runtime-bridge |
Expose global agents.defaults.imageGenerationModel configuration on a developer-only Image Generation page with per-agent auth visibility, independent OpenAI-compatible image endpoint settings, and runtime-backed test generation via Main-process host routes. |
| harness/specs/tasks/image-generation-settings.md |
| electron/utils/openclaw-image-generation-runtime.ts |
| electron/utils/openclaw-image-generation.ts |
| electron/utils/openclaw-image-relay-constants.ts |
| electron/utils/openclaw-auth.ts |
| electron/utils/plugin-install.ts |
| resources/openclaw-plugins/clawx-openai-image/index.mjs |
| resources/openclaw-plugins/clawx-openai-image/openclaw.plugin.json |
| resources/openclaw-plugins/clawx-openai-image/package.json |
| scripts/bundle-openclaw.mjs |
| scripts/patch-openclaw-image-b64-json.mjs |
| package.json |
| electron/services/media-api.ts |
| electron/utils/store.ts |
| electron/services/providers/provider-runtime-sync.ts |
| src/lib/image-generation.ts |
| src/App.tsx |
| src/components/layout/Sidebar.tsx |
| src/components/settings/ImageGenerationSettings.tsx |
| src/pages/ImageGeneration/index.tsx |
| src/pages/Models/index.tsx |
| shared/i18n/locales/*/common.json |
| shared/i18n/locales/*/dashboard.json |
| tests/unit/openclaw-image-generation.test.ts |
| tests/unit/openclaw-auth.test.ts |
| tests/e2e/image-generation-settings.spec.ts |
| tests/e2e/app-smoke.spec.ts |
| README.md |
| README.zh-CN.md |
| README.ja-JP.md |
|
| Models page no longer embeds Image Generation; developer mode shows a dedicated Image Generation sidebar page alongside Skills, Cron, and Dreams. |
| Saving settings writes openclaw.json agents.defaults.imageGenerationModel from the explicit custom image endpoint form; default chat provider changes do not auto-sync image models. |
| The custom image endpoint is always the page's configuration target; no extra enable/disable switch is shown before Base URL/model/API key fields. |
| Saving the OpenAI-compatible image endpoint writes a ClawX-owned provider (`clawx-openai-image`) and auth profile, enables `request.allowPrivateNetwork` for trusted custom endpoints, and leaves `models.providers.openai` untouched so chat continues to use the regular OpenAI provider. |
| Test generate calls OpenClaw in-process generateImage runtime with the selected agent auth directory (no CLI subprocess). |
|
|
| backend-communication-boundary |
| renderer-main-boundary |
| api-client-transport-policy |
|
| tests/unit/openclaw-image-generation.test.ts |
| tests/e2e/image-generation-settings.spec.ts |
|
| Renderer uses typed hostApi media methods only (src/lib/image-generation.ts); no direct Gateway HTTP or ipcRenderer from pages. |
| Image generation settings and test actions are handled in Main process services. |
| Unit tests cover model ref parsing, config read/write, custom endpoint model mapping, private-network endpoint opt-in, and the independent image endpoint not mutating `models.providers.openai`. |
| E2E verifies the Image Generation page is hidden until developer mode is enabled, is not embedded in Models, and exposes the custom endpoint controls. |
|
|