Birdwatcher

AI agent · Data & analytics agent · Data & analytics by Tinybird

An open-source agent for natural language data analysis and monitoring in Tinybird.

Birdwatcher is an open-source AI agent designed to provide data analysis capabilities for data stored in Tinybird. It connects to the Tinybird Model Context Protocol (MCP) server using a secure token to enable users to perform complex analytics via natural language prompts. The agent can be used locally via a CLI or scheduled as a GitHub Action to monitor system health, such as CPU spikes or endpoint errors. It integrates with communication tools like Slack and email to notify teams of its findings. Birdwatcher uses "Missions"—predefined instructions—to guide its analysis, making it a speciali

Languages
English
Deployment options
Local CLI, GitHub Actions
Primary interface
Natural language (Prompt)
Supported integrations
Slack, Resend (Email), Tinybird MCP
Default model support
Claude-4-Sonnet-20250514

Overview

Birdwatcher is an open-source AI agent designed to provide data analysis capabilities for data stored in Tinybird. It connects to the Tinybird Model Context Protocol (MCP) server using a secure token to enable users to perform complex analytics via natural language prompts. The agent can be used locally via a CLI or scheduled as a GitHub Action to monitor system health, such as CPU spikes or endpoint errors. It integrates with communication tools like Slack and email to notify teams of its findings. Birdwatcher uses "Missions"—predefined instructions—to guide its analysis, making it a specialized tool for engineering and data teams.

Data Analysis Capabilities

  • Natural language analytics: Connects to the Tinybird MCP server using a Tinybird token to enable analytics via natural language.
  • Performance investigation: Investigates CPU spikes and other performance anomalies within the last hour.
  • Error monitoring: Monitors and reports endpoint errors occurring in the last 24 hours.
  • Traffic analysis: Analyzes website visits and identifies top pages by traffic volume.

Integrations and Workflow

  • Slack integration: Sends response notifications and analysis results directly to specified Slack channels.
  • Email notifications: Notifies users of analysis findings via email using Resend API keys.
  • GitHub Actions scheduling: Can be scheduled as a GitHub Action to run daily at specific times or on workflow dispatch.
  • Interactive CLI: Includes a Chat CLI mode for interactive data exploration and summaries.

Agent Control and Setup

  • Mission-based guidance: Uses predefined instructions called Missions to guide the agent through specific analysis tasks.
  • Customizable missions: Allows users to provide custom mission rules through the --mission flag for tailored analysis.
  • Model flexibility: Supports model selection including Claude-4-Sonnet for processing prompts.

Specifications

Deployment options
Local CLI, GitHub Actions
Primary interface
Natural language (Prompt)
Supported integrations
Slack, Resend (Email), Tinybird MCP
Default model support
Claude-4-Sonnet-20250514

Added by AntSpace from public sources (tinybird.co).

More AI agents by Tinybird