Overview
This issue tracks the implementation of embedding support in the WordPress AI plugin. Existing embedding work (#683, #891, #943) should be divided into smaller, independently reviewable PRs rather than combining foundational architecture and user-facing features into large feature PRs.
Proposed Work
Each PR should generally:
- Branch independently from the
develop branch
- Be reviewable and mergeable on its own
- Avoid including user-facing features before the required foundation is available
- Include focused tests and documentation
Related PRs may be stacked when a direct implementation dependency makes that more practical.
Testing
Local models, including Ollama, should be supported as a practical way to test embedding workflows without ongoing provider API costs.
Embeddings generally only need to be regenerated when their source content changes. Similarity calculations and other operations using stored embeddings should not require additional provider requests.
Links to the relevant issues and PRs should be added to the checklist above as the work is defined.
Overview
This issue tracks the implementation of embedding support in the WordPress AI plugin. Existing embedding work (#683, #891, #943) should be divided into smaller, independently reviewable PRs rather than combining foundational architecture and user-facing features into large feature PRs.
Proposed Work
Each PR should generally:
developbranchRelated PRs may be stacked when a direct implementation dependency makes that more practical.
Testing
Local models, including Ollama, should be supported as a practical way to test embedding workflows without ongoing provider API costs.
Embeddings generally only need to be regenerated when their source content changes. Similarity calculations and other operations using stored embeddings should not require additional provider requests.
Links to the relevant issues and PRs should be added to the checklist above as the work is defined.