TinySearch

v0.1.0 Feature

This release adds 2 notable features for engineering teams evaluating rollout.

Published 20d Search Engines

View tool

✓ No known CVEs patched

Read the diff → Tool health → What is this tool? →

✓ No known CVEs patched in this version

ReleasePort's take

Light signal

editorial:auto 13d

TinySearch v0.1.0 introduces optional ONNX bundles for faster cold starts and a new core architecture implementing DuckDuckGo-style search with dense + BM25 fusion.

Why it matters: Test the ONNX embeddings in dev to evaluate cold-start latency improvements; assess the new search architecture for relevance before production deployment.

Summary

AI summary

First public release of TinySearch with faster ONNX embeddings.

Changes in this release

Type	Severity	Summary	CVE
Feature
Feature	Medium	Optional ONNX bundle for all-MiniLM-L6-v2 embeddings speeds up cold starts. Optional ONNX bundle for all-MiniLM-L6-v2 embeddings speeds up cold starts. Source: llm_adapter@2026-05-21 Confidence: low	—
Feature	Medium	Core architecture implements DuckDuckGo search, dense + BM25 fusion, crawl, chunking, global reranking, and SEARCH-GROUNDED ANSWER PROMPT. Core architecture implements DuckDuckGo search, dense + BM25 fusion, crawl, chunking, global reranking, and SEARCH-GROUNDED ANSWER PROMPT. Source: llm_adapter@2026-05-21 Confidence: low	—
Feature	Medium	MCP server integration included via `servers/mcp_server.py`. MCP server integration included via `servers/mcp_server.py`. Source: llm_adapter@2026-05-21 Confidence: low	—
Feature	Medium	Optional HTTP API provided by `servers/fastapi_server.py`. Optional HTTP API provided by `servers/fastapi_server.py`. Source: llm_adapter@2026-05-21 Confidence: low	—
Performance	Medium	ONNX embedding bundle accelerates cold starts, especially in MCP subprocesses. ONNX embedding bundle accelerates cold starts, especially in MCP subprocesses. Source: granite4.1:30b@2026-05-22-audit Confidence: low	—

Full changelog

TinySearch v0.1.0

First public release.

Highlights

Faster embeddings: optional embedded ONNX bundle for all-MiniLM-L6-v2 (via onnxruntime) for quicker cold starts, especially in MCP subprocesses — see models/all-minilm-l6-v2-onnx/ and scripts/export_embedding_onnx.py.
Core architecture: DuckDuckGo search → dense + BM25 fusion → crawl → chunk → global rerank → SEARCH-GROUNDED ANSWER PROMPT for your model.
Agent integration: MCP server (servers/mcp_server.py) and optional HTTP API (servers/fastapi_server.py).

Install

See the README for venv setup, MCP config, and environment variables (TINYSEARCH_HF_CACHE, TINYSEARCH_ONNX_MODEL_DIR).

View diff on GitHub

Weekly OSS security release digest.

The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.

No spam, unsubscribe anytime.

Share this release

Share on X Share on Bluesky

Track TinySearch

Get notified when new releases ship.

About TinySearch

All releases →