AI Tools Weekly Sage logoAI Tools WeeklySage
ai-newsnews

Agtop: Btop but for Your Agents

Agtop: Btop but for Your Agents

4 min readAI Tools Weekly
Disclosure: This article contains affiliate links. We earn a commission if you purchase through our links, at no extra cost to you.

Article Title: AI Tools of the Week: Agtop and Claude Code Routines

Lead Story: Agtop - Your AI Coding Agents' Dashboard

Today, we dive into two groundbreaking tools that are transforming how developers interact with their AI agents. First up is agtop, a terminal dashboard designed to give developers unparalleled visibility into their Claude Code and Codex sessions. With agtop, you can track every session's performance in real-time, monitor token usage, and even see which models your agents have processed.

But it's not just for monitoring; agtop also offers features like cost tracking, context pressure analysis, and detailed process logs. For example, on a macOS system, you can easily view CPU and memory metrics for each running session, providing insights into where optimizations might be needed. The npx command at the top of your terminal makes setting up agtop incredibly easy—just run npx @ldegio/agtop to start tracking sessions automatically.

agtop also supports non-running sessions through an inactivity filter, allowing you to clean up unused resources without any hassle. Plus, it provides detailed configuration panels for browsing session data and managing tool activities. Whether you're a seasoned developer or new to Claude Code, agtop offers a user-friendly interface that simplifies access to complex AI agent metrics.

For more details on how agtop works, including its full list of features and configurations, visit the GitHub repository: https://github.com/ldegio/agtop.


What Else Happened Today

In addition to the groundbreaking dashboard tool, today we also learned about Claude Code's new feature: routines. Routines are automated tasks that can run on a schedule or in response to specific events, such as API calls or GitHub triggers. For developers building with Claude Code, this opens up new possibilities for streamlining workflows and reducing human intervention.

For example, teams can set up scheduled routines to handle bug fixes nightly or automate infrastructure updates weekly. They can also create API-triggered routines that respond to events like PRs or security vulnerabilities, sending automated triage summaries to internal channels. Additionally, Claude Code's web-based platform allows for webhook triggers, enabling real-time session management based on user actions.

Routines come with built-in access to your repositories and connectors, making it easy to package automations into self-contained endpoints. However, there are limits on routine usage based on subscription plans: Pro users can run up to 5 routines per day, while Max, Team, and Enterprise users can handle up to 25, 25, and unlimited respectively.


Why This Matters

The introduction of agtop and Claude Code's routines represent significant advancements in how developers interact with AI agents. Agtop simplifies monitoring and managing AI activity, which is crucial as the complexity of these systems grows. Meanwhile, routines offer a scalable way to automate tasks without requiring constant human oversight—potentially revolutionizing development workflows across industries.

For enterprises relying on Claude Code for their CI/CD pipelines or software development cycles, these tools could significantly reduce operational overhead and improve efficiency. Developers will likely adopt them to streamline processes like bug triage, deployment verification, and knowledge sharing within teams.


What to Watch Next

As we approach the end of the week, keep an eye on Claude Code's expanding capabilities, particularly in integrating with external systems beyond GitHub. Additionally, stay tuned for updates to agtop that could enhance its real-time monitoring features or expand its supported platforms. Both tools are poised to become indispensable for developers working with AI agents, and their continued evolution will undoubtedly shape the future of automated development practices.


Sources