Introducing Sirchmunk v0.0.1 — Initial Release

We are excited to announce the initial release of Sirchmunk — an open-source, embedding-free, agentic search engine that transforms raw data into self-evolving intelligence in real time.

What is Sirchmunk?

Sirchmunk takes a fundamentally different approach to retrieval-augmented generation. Instead of relying on static vector embeddings and pre-computed indexes, it operates directly on raw files — delivering instant, full-fidelity retrieval without any pre-processing.

Key Features in v0.0.1

Indexless Retrieval

Drop your files and search immediately. No vector database, no ETL pipeline, no pre-indexing required. Sirchmunk leverages ripgrep-all to search across 100+ file formats in a streaming fashion.

Self-Evolving Knowledge Clusters

Knowledge compounds with every search. The system builds structured Knowledge Clusters that evolve from Emerging to Stable as they are validated across multiple queries.

Monte Carlo Evidence Sampling

Instead of reading entire documents, Sirchmunk uses an exploration-exploitation approach to strategically sample and score document regions — extracting precise evidence while minimizing LLM token costs.

ReAct Agent Fallback

When standard retrieval falls short, an autonomous ReAct agent takes over with iterative reasoning-and-action cycles, exploring alternative strategies until answers are found.

Web UI

A modern web interface with real-time streaming chat, source citations, knowledge cluster browsing, and system monitoring dashboards. Supports dark/light themes and bilingual localization.

Python SDK

import asyncio
from sirchmunk import AgenticSearch
from sirchmunk.llm import OpenAIChat

llm = OpenAIChat(
    api_key="your-api-key",
    base_url="your-base-url",
    model="your-model-name"
)

async def main():
    searcher = AgenticSearch(llm=llm)
    result = await searcher.search(
        query="How does transformer attention work?",
        paths=["/path/to/documents"],
    )
    print(result)

asyncio.run(main())

Getting Started

pip install sirchmunk

For the full documentation, see the Getting Started guide.

What’s Next?

We are actively working on MCP integration, CLI improvements, and knowledge persistence. Stay tuned for v0.0.2!

GitHub Repository · ModelScope

No results found