Wire Prompt Garden imports into workspace and favorite flows, add the import dialog/panel affordances, persist Garden metadata/media into favorites, and document the external import contract.
- Pass existing filled variable values into smart value generation
- Update generation templates and service contracts for context-aware output
- Cover service runtime and UI composable behavior with targeted tests
- Bind favorites to source assets and save targets throughout workspace actions
- Share the favorite editor/detail surfaces across dialogs and workspace panels
- Add provenance badges, utility-menu apply actions, and themed favorite panels
- Expand unit and integration coverage for favorite source and editor flows
- Add prompt-model services for content, examples, modes, sessions, and history
- Persist richer favorite prompt-asset metadata and session projections
- Wire favorite save/apply paths across supported workspaces
- Cover the new prompt-asset contract with core and UI regressions
- strengthen locale parity checks and related regression coverage
- stabilize repo-level english-first test expectations
- harden MCP packaging, install entrypoints, and supporting script validation
- tighten related config and model checks used by the rollout
- extract evaluation, rewrite, and structured-compare template content into dedicated assets
- separate builder wiring from runtime copy and content definitions
- simplify evaluation service assembly around shared template modules
- add guardrail coverage for extracted evaluation templates
- split UI locale packs into en-US, zh-CN, and zh-TW modules
- switch runtime defaults, docs, and repository guidance toward an English-first baseline
- update core, MCP, and UI runtime copy to remove hardcoded Chinese fallbacks
- add locale parity and no-hardcoded-runtime guardrails across the repo
- include ordered OpenRouter multi-image parts in provider payloads
- persist multi-image session inputs as asset references instead of inline payloads
- restore ordered inputs from storage and cover the persistence path with tests
- Strengthen the style migration contract around natural subject integration and visual hierarchy
- Remove the image2image reference extraction entry from the migrated UI flow
- Normalize migration placeholders and document the updated migration design
- Align Cloudflare model defaults, metadata, and connection labels across core and UI
- Add transient retry handling for Cloudflare image requests
- Harden reference prompt normalization and extend related regression coverage
- replace the direct extract-from-image path with a reference dialog preview and apply workflow
- centralize reference prompt preview state and temporary variable syncing inside the text2image workspace
- preserve text2image evaluation wiring and compare or result review actions while refactoring the reference flow
- Backfill builtin model API keys when restoring stored model configurations
- Keep model manager defaults aligned with the new image model loading flow
- Add focused manager test coverage for missing-key recovery
- add a Workers AI image adapter for FLUX.2 [klein] 4B generation and editing
- add a Cloudflare text adapter with dynamic model discovery and a Qwen3 default
- load Cloudflare token and account defaults from environment variables for both providers
- cover registry integration and default config wiring with unit tests
(cherry picked from commit 3ef9bdbd187ef150d331ef6b9b46f9d7ad3ef2af)
- Add MiniMax-M2.7 and MiniMax-M2.7-highspeed to model list
- Set MiniMax-M2.7 as default model (first in list)
- Keep all previous M2.5 models as alternatives
- Update related unit tests
Co-authored-by: PR Bot <pr-bot@minimaxi.com>
- expand core and ui regression coverage for analysis, result, and compare evaluation flows
- refresh text analysis and evaluation VCR fixtures after the workspace/result semantics change
- cover stale-state and analyze-created workspace behaviors in integration and e2e tests
- replace original/optimized semantics with analysis, result, and compare request types
- centralize evaluation prompt construction in shared builders
- tighten rendered sample and evidence behavior tests for core evaluation service
