A comprehensive observability tool for tracking AI agent API calls, error rates, and system resource utilization.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install agent-metrics-osiris
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install agent-metrics-osiris using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Agent Metrics?
The Agent Metrics skill provides developers with a robust framework for monitoring AI agent behavior and performance in real-time. By integrating this tool into your workflow, you can capture critical data points such as API call counts, response latencies, and detailed error logs. This skill is a vital component of the Openclaw Skills library, designed to help developers maintain high reliability and performance standards for their automated agents through actionable insights and clear data visualization.
Agent Metrics Use Cases
- Debugging agent failures by reviewing logged stack traces and specific error details.
- Benchmarking the performance of different LLM providers by tracking and comparing average latency.
- Monitoring operational costs and rate limits by analyzing the frequency of API calls.
- Visualizing agent health through a terminal-based dashboard during development or production runs.
How Agent Metrics Works
- Initialize the environment by installing the required Python dependencies to handle system-level metrics.
- Trigger record commands via the Python CLI or PowerShell wrapper whenever a relevant agent event, such as an API request or an error, occurs.
- The skill captures the event type, label, and associated metadata (like duration or error messages) into a structured format.
- Aggregate and analyze the data using built-in dashboard, summary, or export actions to identify performance bottlenecks.
Agent Metrics Setup
To begin using this skill from the Openclaw Skills collection, first install the necessary dependencies:
pip install psutil
Verify the installation by recording a test metric using the cross-platform Python CLI:
python metrics.py record --type call --label "initialization_test"
Agent Metrics Data Schema & Taxonomy
The skill organizes monitoring data into four primary categories to ensure clear observability:
| Metric Type | Purpose | Key Data Fields |
|---|---|---|
| call | Tracks usage volume | label, timestamp |
| error | Logs execution failures | label, details, timestamp |
| latency | Measures response speed | label, value (ms), timestamp |
| custom | User-defined tracking | label, value |
Agent Metrics Advanced Features
- Deep system resource monitoring including real-time CPU, Memory, and Disk usage statistics.
- Native JSON export functionality to support external dashboarding and long-term data retention.
- Integrated terminal dashboard providing a high-level overview of error rates, uptime, and top-performing labels.
- Flexible cross-platform support ensuring consistent metrics collection across Windows, macOS, and Linux as part of the Openclaw Skills ecosystem.
SKILL.md
Loading
Related Openclaw Skills
Agent Identity Skill
A robust cryptographic framework for AI agents to sign messages, verify identities, and establish secure agent-to-agent communications.
A2A Protocol Skill
A standardized implementation of the Agent2Agent Protocol enabling discovery, messaging, and task delegation between autonomous AI agents.
Discord Task Center
A professional integration that turns Discord forum threads into structured AI task sessions with automated tagging and model switching.
Discord Dynasty - Task & Server Management
A specialized skill for OpenClaw agents to manage Discord forum-based task centers and deploy historical-themed server hierarchies.
Agent Watcher
A specialized monitoring skill for Moltbook that detects new agents, tracks keyword-specific posts, and synchronizes findings with Open Notebook or local files.
ArXiv Osiris
A research-focused skill for OpenClaw agents to programmatically search, filter, and download scientific papers from the arXiv repository.
Featured*
