Tunnlo
Tunnlo is an open-source, TypeScript-based tool that connects real-time data streams to LLM-powered agents for monitoring, triaging, and action. It offers smart filtering, multi-agent orchestration, per-agent dashboards, and is designed to be self-hosted with privacy-first principles.
Key facts
What is Tunnlo?
Bridge real-time data to intelligent agents that monitor, triage, and act, filtering noise so only what matters reaches the right LLM.
Who is Tunnlo best for?
Engineering teams, DevOps, SRE, Security analysts, Product managers, Data scientists
Key features
- ✓Pluggable adapters for Kafka, Google Docs, network traffic, logs, stdin, MCP servers
- ✓Smart filtering: rate limiting, deduplication, content matching, adaptive sampling, windowed aggregation, priority routing
- ✓Multi-agent & multi-LLM support: run agents in parallel using Anthropic, OpenAI, Ollama, LangGraph, CrewAI, OpenClaw
- ✓Action dispatch via webhooks, MCP tool calls, or approval-gated human-in-the-loop operations
- ✓Built-in observability: per-agent response panels, side-by-side comparison, live metrics, filter drop rates, token usage tracking
- ✓Privacy first: self-hosted, no telemetry, no cloud dependency; runs fully offline with local LLMs
Use cases
- →Intelligent log monitoring: filter log noise, triage errors, escalate only what matters
- →Real-time threat detection: capture network traffic, classify threats, surface real attacks
- →User behavior intelligence: sample high-volume streams, detect churn signals and sentiment shifts
- →AI-powered document collaboration: watch Google Docs, summarize feedback, evaluate suggested edits
- →Multi-agent cross-source correlation: route different data sources to specialized LLM agents and correlate insights
Pricing
Open Source
$0
- · All features including smart filtering, multi-agent, dashboards
- · Self-hosted
- · No telemetry
- · MIT License
Free and open source (MIT License)
Pros
- +Open source with MIT License
- +Fully self-hosted and privacy-first (no telemetry)
- +Supports local LLMs via Ollama for offline use
- +Pluggable architecture for custom data sources and agent runtimes
- +Quick setup with npm and YAML configuration
Positioning
- Core value: Bridge real-time data to intelligent agents that monitor, triage, and act, filtering noise so only what matters reaches the right LLM.
- Ideal for: Engineering teams, DevOps, SRE, Security analysts, Product managers, Data scientists
- Product type: Open-source CLI tool with a web dashboard
Frequently asked questions
What is Tunnlo?
Tunnlo is an open-source tool that connects real-time data streams to LLM-powered agents for monitoring, triage, and action. It uses smart filtering and multi-agent orchestration to surface only what matters.
How do I install Tunnlo?
Install via npm: `npm install -g tunnlo` or create a new project with `npm create tunnlo@latest`. Then run with `npx tunnlo start pipeline.yaml`.
Does Tunnlo support local LLMs?
Yes, Tunnlo integrates with Ollama for local LLMs and also supports Anthropic, OpenAI, LangGraph, CrewAI, and OpenClaw.
Can I self-host Tunnlo?
Yes, Tunnlo is designed to be self-hosted. It has no telemetry, no cloud dependency, and can run fully offline with local LLMs.
Is Tunnlo free?
Yes, Tunnlo is open source under the MIT License and free to use.
What data sources does Tunnlo support?
Tunnlo supports Kafka, Google Docs, network traffic (tshark), log files, stdin, MCP servers, and custom adapters via the Adapter SDK.
You may also like
Curated suggestions based on similar categories.
Agentcard for companies
AgentCard is a virtual card platform for AI agents, enabling them to make online purchases autonomously. Users create prepaid cards, set spending limits, and track all transactions in real time. Currently in private beta, it works with any AI agent and supports merchants like DoorDash, OpenAI, and Amazon.
loopclub
LoopClub.com is a venture intelligence platform that uses AI to curate opportunity pipelines, automate due diligence, and provide portfolio analytics for founders and investors. It is part of the eCorp network and currently offers free early access.
Synthia
SYNTHIA is a synthetic data generation framework for integrated validation of AI healthcare applications. It provides privacy-preserving synthetic data across various healthcare data types to accelerate personalized medicine and research. The project is a multidisciplinary collaboration of 39 partners developing validated tools and methods for synthetic data generation.
Basedash SCIM
Basedash is an AI-powered business intelligence platform that generates trusted charts, dashboards, and reports on a governed semantic layer, supporting cloud or self-hosted deployment and connectivity to 750+ data sources.
Muse Spark 1.1 by Meta AI
Muse Spark is a natively multimodal reasoning model from Meta Superintelligence Labs, the first in the Muse family. It features tool-use, visual chain of thought, and multi-agent orchestration, and is available at meta.ai and the Meta AI app.
SmartRolePath
SmartRolePath is an AI career intelligence platform that analyzes a user's skills against live hiring data from 50,000+ job postings to generate personalized career paths with real salary data, market demand scores, automation risk assessments, and month-by-month transition roadmaps.