llm-search
pyLLMSearch is a Python package for building retrieval-augmented generation (RAG) systems that query local document collections. It provides YAML-based configuration, support for multiple document formats (PDF, Markdown, DOCX), hybrid search with re-ranking, and integrates with various LLMs via OpenAI-compatible APIs or local models.
Key facts
Objective fields from the source. Values we can't verify are shown as “Unknown” rather than guessed.
| Field | Value |
|---|---|
| Repository | snexus/llm-search |
| Owner | snexus |
| Primary language | Jupyter Notebook |
| License | MIT — OSI-approved |
| Stars | 659 |
| Forks | 71 |
| Open issues | 6 |
| Latest release | v0.9.5 (2026-01-17) |
| Last updated | 2026-01-17 |
| Source | https://github.com/snexus/llm-search |
What llm-search is
A RAG framework built on FastAPI and ChromaDB that implements dense + sparse embeddings (SPLADE), HyDE, multi-query generation, cross-encoder re-ranking, and MCP server support. Supports custom embedding models (HuggingFace, Sentence-Transformers, OpenAI) and LiteLLM for model interoperability.
Get the llm-search source
Clone the repository and explore it locally.
git clone https://github.com/snexus/llm-search.gitcd llm-search# follow the project's README for install & configurationNeed it deployed, integrated, or customized instead? DEV.co ships production installs.
Best use cases
Implementation considerations
- Requires Python environment setup and familiarity with YAML configuration; evaluate learning curve for team skill level.
- ChromaDB setup for vector storage and model hosting (OpenAI API or local LLM) must be planned and resourced.
- Re-ranking and HyDE are powerful but optional; enable incrementally and validate quality trade-offs against latency.
- Document parsing quality varies by format; test with representative samples (PDFs, tables, images) before large-scale indexing.
- MCP server integration requires compatible client setup (Cursor, Windsurf, etc.); not all editor/IDE combinations tested.
When to avoid it — and what to weigh
- Real-time Streaming at Massive Scale — The package is optimized for medium-size document bases; no data provided on scaling behavior for terabyte-scale collections or sub-second latency requirements.
- Regulatory Compliance Without Audit Trail — No explicit logging, audit, or compliance-focused features documented. Organizations requiring HIPAA, SOC 2, or similar certifications should conduct thorough review.
- Turnkey Commercial Product — This is a library requiring custom integration, configuration, and deployment; it is not a managed SaaS. Teams without ML/DevOps expertise may face operational burden.
- Latency-Sensitive Applications — Multi-stage retrieval (dense + sparse + re-ranking + optional HyDE) may introduce latency; no benchmarks provided for response time constraints.
License & commercial use
MIT License. Permissive open-source license allowing commercial use, modification, and distribution with attribution.
MIT is a permissive OSI license compatible with commercial deployment. However, review dependencies (ChromaDB, Unstructured, LiteLLM, SPLADE, model licenses) for compliance. No explicit commercial support or indemnification from the maintainer; you assume all liability for your integration.
DEV.co evaluation signals
Editorial assessment — not user reviews. Directional, with an explicit confidence level.
| Signal | Assessment |
|---|---|
| Maintenance | Active |
| Documentation | Adequate |
| License clarity | Clear |
| Deployment complexity | High |
| DEV.co fit | Good |
| Assessment confidence | High |
Local document processing mitigates data exfiltration risk if using on-premise LLMs, but: (1) no explicit data encryption, authentication, or role-based access controls documented; (2) if using OpenAI/external APIs, review model endpoint exposure and rate-limit mechanisms; (3) MCP server interface should be firewalled in shared environments; (4) dependency scanning and supply-chain risk assessment recommended before production use.
Alternatives to consider
LangChain + LlamaIndex
Mature, widely-adopted frameworks with larger ecosystems and community support. More opinionated but potentially steeper learning curve and less control over search strategy.
Verba by Weaviate
Full-stack RAG UI with Weaviate vector DB and modular architecture. More turnkey; less direct control over parsing and re-ranking pipelines.
Production-grade RAG framework with serializable pipelines and extensive integrations. Steeper setup overhead; better for enterprise workflows.
Build on llm-search with DEV.co software developers
Evaluate pyLLMSearch for your document Q&A needs. We help integrate, deploy, and scale RAG systems with custom LLMs and enterprise compliance. Let's discuss your architecture.
Talk to DEV.coRelated open-source tools
Surfaced by semantic similarity across the DEV.co open-source index.
Related on DEV.co
Explore the category and the services that help you build with it.
llm-search FAQ
Can I use my own LLM without OpenAI?
What is HyDE and when should I enable it?
How large can my document base be?
Does it support image parsing?
Software development & web development with DEV.co
Need help beyond evaluating llm-search? DEV.co is a software development agency offering software development services and web development for teams of every size. Our software developers and web developers build custom software, web applications, APIs, and rag frameworks integrations — and maintain them long-term.
Ready to Build Your RAG System?
Evaluate pyLLMSearch for your document Q&A needs. We help integrate, deploy, and scale RAG systems with custom LLMs and enterprise compliance. Let's discuss your architecture.