Weave Code
Vector Laravel Package
the-shit/vector
A lightweight Laravel/PHP package for vector math and data handling, offering a simple API for creating, manipulating, and integrating vectors into your applications. Designed for developers who need straightforward vector utilities without heavy dependencies.
the-shit/vector is a lightweight Laravel/PHP package for working with vectors and similarity search. It offers a clean, expressive API for representing vectors, calculating distances, and adding semantic ranking to Laravel applications without heavy dependencies.
- Simple Vector value object and fluent helpers
- Cosine similarity, Euclidean distance, and normalization
- Laravel-friendly integration and configuration
- Useful for embeddings and semantic search workflows
- Extensible, testable design for custom metrics
Frequently asked questions about Vector
- How do I install the-shit/vector in a Laravel application?
- Install it with Composer using the package name the-shit/vector. Because it is a Saloon-based Qdrant connector, make sure Saloon is available if Composer does not already pull it in. Check the package README for any publishable configuration, environment variables, or migration steps needed for your Qdrant connection and embedding storage.
- Which Laravel versions does the-shit/vector support?
- The exact supported Laravel and PHP versions are defined in the package composer.json constraints. Check those constraints before installing it into an older or very new Laravel application. If the repository or README does not list compatibility, use Composer to verify requirements for your project.
- How do I attach vector embeddings to an Eloquent model?
- The package is designed to store and query vector embeddings alongside Eloquent models. You typically add the package's trait or interface to a model, configure which fields are embedded, and persist the generated vector with the record. Confirm the exact trait, interface, and column names in the README because those API details are package-specific.
- Do I need a running Qdrant server to use this package?
- Yes. The package is a thin Saloon connector for Qdrant, so it needs a reachable Qdrant instance and the correct URL and API key configuration. You can run Qdrant locally with Docker for development or connect to a managed Qdrant deployment in production. Without Qdrant, the similarity search and vector persistence features cannot work.
- How are embeddings generated and stored with this Laravel package?
- The package provides embedding generation so vectors can be created and persisted alongside your Eloquent data. You configure your embedding provider credentials and model according to the README, then trigger generation for the fields you want to index. The resulting vectors are sent to Qdrant for storage and later similarity queries.
- Can I use the-shit/vector for semantic search and recommendations?
- Yes. Its main use case is running vector similarity searches against Qdrant to power semantic search, recommendations, and other AI-driven features. You query with a vector and receive the nearest matches from your Qdrant collection. Qdrant payload filters can help constrain results by model type, tenant, or other metadata.
- Is the-shit/vector suitable for production use?
- It can be used in production, but it is a thin connector, so reliability depends heavily on your Qdrant deployment, batching strategy, timeouts, and retry configuration. Embedding generation is often slow, so run it in queued jobs rather than synchronous web requests. Also monitor Qdrant latency, collection indexing, and API error rates before scaling to large datasets.
- How can I test the Qdrant connector without hitting a real Qdrant instance?
- Because the package uses Saloon, you can typically mock connector requests with Saloon's mocking utilities instead of calling a live Qdrant server. Unit test your embedding and model logic separately from the HTTP layer. For integration confidence, run Qdrant in a container during CI and test real collection creation, upserts, and searches.
- What are alternatives to the-shit/vector for vector search in Laravel?
- Alternatives include using the official Qdrant client directly, Laravel Scout with a vector-capable driver, or packages for other vector stores such as pgvector, Pinecone, or Weaviate. This package is a good fit if you want a thin, composable Saloon connector and already use Qdrant. Compare based on your existing database, hosting, filtering needs, and whether you want Eloquent integration.
- Can I customize the Saloon requests or extend the connector?
- The package is described as thin and composable, so it is intended to let you extend or replace Saloon requests and connector behavior where needed. You can add custom endpoints, payloads, or request middleware by following Saloon patterns. Check the package source and README for the exact extension points before modifying vendor code.
Weaver
How can I help you explore Laravel packages today?