Add ClickHouse vector store provider

[MitraShabani]

Implements the VectorStoreBase interface for ClickHouse, following the pattern used by the existing Qdrant/Chroma providers.

• Add ClickhouseDB in mem0/vector_stores/clickhouse.py
• Add ClickhouseConfig in mem0/configs/vector_stores/clickhouse.py
• Register clickhouse in factory.py and vector_stores/configs.py
• Add clickhouse-connect as an optional dependency
• Add tests (tests/vector_stores/test_clickhouse.py)
• Add docs page and overview.mdx entry

Linked Issue

Closes #6991

Description

Mem0 currently lacks support for ClickHouse as a vector store backend. This PR adds a ClickhouseDB provider implementing the full VectorStoreBase interface (create_col, insert, search, delete, update, get, list_cols, delete_col, col_info, list, reset), so users already running ClickHouse for other workloads can use it as their memory backend without standing up a separate vector database.

Similarity search uses ClickHouse's built-in cosineDistance() function, converted to a higher-is-better similarity score as required by the base interface. Deletes and upserts are implemented via ALTER TABLE ... DELETE followed by insert, since ClickHouse's MergeTree engine has no native in-place update.

Type of Change

• Bug fix (non-breaking change that fixes an issue)
• New feature (non-breaking change that adds functionality)
• Breaking change (fix or feature that would cause existing functionality to change)
• Refactor (no functional changes)
• Documentation update

AI Assistance

• No AI assistance
• AI-assisted (autocomplete, or I asked a model questions while writing this)
• AI-generated (an agent wrote most or all of this diff)

Used Claude interactively throughout — to understand the VectorStoreBase interface and existing provider patterns, to work through implementing and debugging each method, and to diagnose issues (e.g. an async-delete timing race, a dict-vs-object mismatch with mem0's internal result handling). I wrote and tested every method myself against a local ClickHouse instance and understand what each part does and why.

• I can explain every line of this diff and how it interacts with the rest of the codebase, without asking an AI tool.

Breaking Changes

N/A — this adds a new optional provider and does not change any existing behavior or interfaces.

Test Coverage

• I added/updated unit tests
• I added/updated integration tests
• I tested manually (describe below)
• No tests needed (explain why)

Added tests/vector_stores/test_clickhouse.py with 9 tests covering collection creation, insert/get, upsert-not-duplicate behavior, search ranking, delete, update, col_info, and reset. Tests skip automatically if ClickHouse isn't reachable, following the pattern in test_pgvector.py.

Also manually verified end-to-end through mem0's public API (Memory.from_config(...) with provider: "clickhouse"), using real OpenAI embeddings for both add() and search() — not just the isolated ClickhouseDB class.

Checklist

• My code follows the project's style guidelines
• I have performed a self-review of my code
• I have added tests that prove my fix/feature works
• New and existing tests pass locally
• I have updated documentation if needed

[CLAassistant]

All committers have signed the CLA.

[github-actions]

Hi @MitraShabani, thanks for opening this pull request.

This is just a soft check: you are not yet in this repo's vouched contributor list (.github/VOUCHED.td). Nothing is blocked and there is nothing you need to do.

A maintainer can vouch for you by commenting !vouch @MitraShabani on any issue.