feat: support configurable embedding dimensions via VECTOR_STORE_DIMENSIONS

[TreyDong]

Summary

Currently embedding_client.py and models.py hardcode 1536 as the embedding dimension. This prevents using embedding providers with different dimensions (e.g., BGE-M3 at 1024 dimensions).

Changes

Replace hardcoded 1536 with settings.VECTOR_STORE.DIMENSIONS in:

• embedding_client.py: 3 Gemini embed_content calls (lines ~85, ~119, ~255)
• models.py: 2 Vector() columns — MessageEmbedding.embedding and Document.embedding — and add from .config import settings

Motivation

The VECTOR_STORE_DIMENSIONS config field already existed in src/config.py but was not being used in embedding_client.py or models.py. This change wires it up, allowing users to configure embedding dimensions via the VECTOR_STORE_DIMENSIONS environment variable instead of being locked to 1536.

This is particularly useful for embedding providers like BGE-M3 (1024 dimensions).

Backwards Compatibility

Default remains 1536 when the VECTOR_STORE_DIMENSIONS env var is unset.

Issue

Closes #463

Summary by CodeRabbit

• Chores
  • Updated embedding vector configuration to use centralized settings instead of hardcoded values, improving system flexibility and maintainability.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: e84ef657-4ff7-4052-bdc8-16ba7035dbdf

📥 Commits

Reviewing files that changed from the base of the PR and between a5423b5 and 82a08e9.

📒 Files selected for processing (2)

• src/embedding_client.py
• src/models.py

---

Walkthrough

The changes replace hardcoded embedding dimension values of 1536 with the configurable settings.VECTOR_STORE.DIMENSIONS setting across embedding client request configuration and vector column definitions in data models.

Changes

Cohort / File(s)	Summary
Embedding Client Configuration src/embedding_client.py	Replaced hardcoded output_dimensionality: 1536 with settings.VECTOR_STORE.DIMENSIONS in three embedding request call sites: embed(), simple_batch_embed(), and _process_batch() methods.
Database Model Definitions src/models.py	Added settings import from .config and updated vector column definitions in MessageEmbedding and Document classes from Vector(1536) to Vector(settings.VECTOR_STORE.DIMENSIONS).

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~5 minutes

Poem

🐰 A rabbit hops with glee,
No more hardcoded 1536!
Dimensions flex, now configured free,
BGE-M3's 1024 won't be hexed. ✨

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 66.67% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately and specifically describes the main change: replacing hardcoded embedding dimensions with a configurable settings value.
Linked Issues check	✅ Passed	All requirements from issue #463 are met: hardcoded 1536 values in embedding_client.py (3 locations) and models.py (2 locations) are replaced with settings.VECTOR_STORE.DIMENSIONS, and settings import is added to models.py.
Out of Scope Changes check	✅ Passed	All changes are directly related to issue #463 and the stated objective of replacing hardcoded embedding dimensions with configurable settings values.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[VVoruganti]

This would require additional changes to the alembic migrations and models.py file to support storing different sized embeddings.

Please also update the alembic migrations and sqlalchemy files accordingly