feat(community): add LogosDB semantic retriever tool

[jose-compu]

Summary

Adds LogosDB as a community retriever tool for Cohere Toolkit.

LogosDB is a local, zero-infrastructure HNSW vector database (mmap-backed).
This tool lets Command R+ retrieve from a local semantic index with no
external services — just a directory on disk.

How it works

1. Documents are embedded with Cohere's Embed API (embed-english-v3.0 by default)
2. Vectors are stored in a local HNSW index via LogosDB
3. At query time the user's question is embedded and nearest neighbours are returned as tool results

Configuration

Set these environment variables (or add to configuration.yaml):

Variable	Default	Description
LOGOSDB_PATH	(required)	Root directory for the index
LOGOSDB_EMBED_MODEL	embed-english-v3.0	Cohere embedding model
LOGOSDB_NAMESPACE	default	Collection name

The tool is unavailable (and silently skipped) if LOGOSDB_PATH is not set.

Changes

File	Change
src/community/tools/logosdb.py	New — LogosDBRetriever implementing BaseTool
src/community/tools/__init__.py	Export LogosDBRetriever
src/community/config/tools.py	Register LogosDB in CommunityTool enum
src/backend/config/settings.py	Add LogosDBSettings + wire into ToolSettings
pyproject.toml	Add logosdb >=0.7.3 to community deps group

Links

• PyPI: https://pypi.org/project/logosdb/
• GitHub: https://github.com/jose-compu/logosdb

[cla-assistant]

All committers have signed the CLA.

[cla-assistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}