watchtower

Automate scheduled web surveillance sweeps and generate cited Markdown digests.

86|42|Updated May 6, 2026

Search Tags:#automation #research #monitoring #intelligence #tavily #web-surveillance

Ready to run

npx skills add https://github.com/NVIDIA/nemoclaw-community --skill watchtower

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: watchtower
Source: https://github.com/NVIDIA/nemoclaw-community/tree/main/examples/recipes/partners/tavily/watchtower/skills/watchtower
Command: npx skills add https://github.com/NVIDIA/nemoclaw-community --skill watchtower

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

This skill solves the problem of information overload by automating the monitoring of specific topics across the web, ensuring you only spend time reviewing genuinely new and relevant updates.

Core Features & Use Cases

Scheduled Surveillance: Automatically sweeps a watchlist of topics using web search and extraction tools.
Deterministic Filtering: Uses mechanical deduplication and domain-based exclusion to filter out noise and previously seen content.
Structured Reporting: Generates cited Markdown digests and JSON changelogs for easy integration into your workflow.

Quick Start

Ask the agent to run a watchtower sweep for the active watchlist to monitor for new updates.

Frequently Asked Questions about watchtower

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I automate web surveillance to monitor specific topics for new information?▼

Web surveillance automation works by scheduling sweeps against a watchlist of topics, using search and extraction tools to find new information, filtering out previously seen content, and generating cited Markdown digests and JSON changelogs.

What is the best way to filter out duplicate results when monitoring a topic watchlist?▼

Filtering duplicate results from a watchlist is handled by deterministic filtering, which applies mechanical deduplication and domain-based exclusion to remove previously seen content and noise.

Do I need Tavily search tools to run scheduled monitoring sweeps?▼

Yes, Tavily search and extraction tools are required to maintain the crash-safe, stateful monitoring pipeline that performs the scheduled sweeps and extracts new information.

How does a stateful monitoring pipeline handle crash recovery during a web sweep?▼

A crash-safe, stateful monitoring pipeline maintains historical state to filter new search candidates against previously seen content, ensuring that interrupted runs do not duplicate digested information in later sweeps.

What output formats are generated from automated web intelligence digests?▼

Automated web intelligence digests generate cited Markdown reports and JSON changelogs as output formats, providing structured results for easy integration into editorial review workflows.