TinySearch v0.4.1

What's Changed

• Project repository moved to the TinySuiteHQ GitHub org — all package metadata (server.json, pyproject.toml, Dockerfile, README) now references github.com/TinySuiteHQ/TinySearch.
• CI: removed a redundant push trigger from the Docker publish workflow and added a wait step so the MCP registry publish step doesn't race the Docker image becoming available.

Full Changelog: v0.4.0...v0.4.1

TinySearch v0.4.0

TinySearch 0.4.0 makes local-first, token-efficient web research easier to
install, run, and integrate. Search, crawling, embeddings, and hybrid reranking
turn webpages into compact, source-grounded evidence without requiring a paid
search API.

What changed

• Installable package: TinySearch is now published as tinysuite-search
  with a proper Python API and optional [server] dependencies.
• One-command MCP: launch over stdio with uvx, or use
  tinysearch serve for Streamable HTTP.
• Setup tooling: tinysearch setup prepares Chromium and the local ONNX
  model; tinysearch doctor checks runtime readiness.
• Unified configuration: one validated tinysearch_config.json controls
  search backends, embeddings, crawling, concurrency, ranking, and token
  budgets.
• Structured evidence: Python, MCP, and FastAPI now share the same research
  and scraping pipelines, source attribution, and stable result schema.
• Better retrieval: search-result ranking, cross-source chunk ranking,
  deduplication, and source quotas keep evidence focused and reduce context.
• Flexible search: use DDGS by default, self-hosted SearXNG, or an optional
  Brave fallback.
• Stronger releases: wheels and Docker images are smoke-tested across
  Python 3.12–3.14, with patched pypdf, MCP, and Crawl4AI dependencies.

Upgrade notes

The package is now named tinysuite-search, server dependencies use the
[server] extra, and the primary configuration file is
tinysearch_config.json. Refresh cached MCP tool schemas after upgrading.

Full changelog: v0.3.2...v0.4.0

TinySearch v0.3.2

Full Changelog: v0.3.1...v0.3.2

TinySearch v0.3.1

What's Changed

• Harden scraping and parsing against ReDoS and bump vulnerable dependendencies by @MarcellM01 in #17

Full Changelog: v0.3.0...v0.3.1

v0.3.0

TinySearch v0.3.0

Focused release for agent workflows: read a known URL directly, and fetch the current UTC date/time on demand.

What's new

Direct URL inspection - scrape_url / POST /scrape

When you already have a URL, skip full web search. TinySearch crawls the page, ranks relevant chunks against your query, and returns a URL-grounded prompt in answer.

• MCP: scrape_url(url, query)
• HTTP: POST /scrape

By @benmaster82 in #14

Current datetime - get_current_datetime / GET /current_datetime

On-demand UTC clock for time-sensitive or relative-date questions (latest, this year, last month). Works reliably over stdio, where MCP instructions are only sent at initialize.

• MCP: get_current_datetime() - no parameters
• HTTP: GET /current_datetime

{ "date_utc": "2026-06-28", "time_utc": "08:10:00" }

By @MarcellM01

MCP tools

Tool	Use when
get_current_datetime()	You need the current UTC date/time before research
research(query)	Open-ended discovery - URL not known yet
scrape_url(url, query)	A specific URL is already known

Upgrade notes

No config changes. Existing research(query) behavior is unchanged.

Full Changelog: v0.2.1...v0.3.0

v0.2.1 - Beta

TinySearch v0.2.1 - Beta

Browser-friendly MCP and a simpler Docker quickstart.

Highlights

• CORS for Streamable HTTP MCP: fixes browser clients (e.g. llama.cpp web UI) failing on OPTIONS /mcp preflight
• Docker Compose quickstart: run TinySearch + SearXNG with one command, no clone required (compose.quickstart.yaml)

Docker

docker compose -f "https://github.com/MarcellM01/TinySearch.git#main:compose.quickstart.yaml" pull
docker compose -f "https://github.com/MarcellM01/TinySearch.git#main:compose.quickstart.yaml" up -d

MCP endpoint: http://localhost:8000/mcp

If using a browser UI from another host/IP, point MCP at that host (e.g. http://<host>:8000/mcp), not localhost.

Full Changelog: v0.2.0...v0.2.1

TinySearch v0.2.0 - Beta

TinySearch v0.2.0 - Beta

This release moves TinySearch from the early alpha stage toward a more stable beta.

The main improvement is a new SearXNG search backend, with DuckDuckGo kept as a fallback. This should make search more reliable and configurable, especially after the recent DuckDuckGo availability issues.

What's Changed

• Added SearXNG search backend support
• Added DuckDuckGo fallback behavior when SearXNG is unavailable
• Restored working search flow after recent upstream search reliability issues
• Improved the foundation for configurable search backends going forward

Huge thanks to @benmaster82 for the high-quality implementation in:
#8

Notes

This is not a full 1.0.0 release yet. The project is now moving into beta, meaning the core search flow should be usable again, but configuration, backend support, and MCP integration may still evolve.

Full Changelog: v0.1.4...v0.2.0

TinySearch v0.1.4

What's Changed

• UPDATE: Add configurable global timeout to agentic_run pipeline by @benmaster82 in #5
• UPDATE: Implement blocked domains feature to filter search results by @MarcellM01 in #7

New Contributors

• @benmaster82 made their first contribution in #5. Big thank you from all of us at TinySearch 🥳🎉

Full Changelog: v0.1.3...v0.1.4

TinySearch v0.1.3

What's Changed

• Add multi-arch Docker publish workflow by @MarcellM01 in #3

New Contributors

• @MarcellM01 made their first contribution in #3

Full Changelog: v0.1.2...v0.1.3

TinySearch v0.1.2

v0.1.2

TinySearch now has a much cleaner local embedding and Docker story, with MCP as the first-class runtime.

Highlights

• Added MCP-first Docker support.
• Added Docker Hub-ready image metadata and runtime layout.
• Added Streamable HTTP/SSE-friendly MCP server configuration.
• Added multiple built-in ONNX embedding presets.
• Added custom Hugging Face ONNX model support.
• Removed sentence-transformers from runtime dependencies.
• Cleaned up local model storage so model weights stay out of Git.

Docker

TinySearch can now run as a Dockerized MCP server.

Published image:

docker pull marcellm01/tinysearch:v0.1.2