How does it help with debugging failed AI agents?

By categorizing errors into types like timeouts, validation errors, or auth failures, it allows developers to quickly pinpoint why an agent's tool call failed.

What is Tool Call Tracking for Claude Code?

It is a specialized skill that instruments tool and function executions to monitor how AI agents interact with external systems, providing data on success, failure, and performance.

What metrics are tracked for each tool execution?

It tracks essential attributes including tool name, success status, latency in milliseconds, argument summaries, and result metadata.

Can I use this with frameworks like LangChain or CrewAI?

Yes, the skill provides specific integration patterns, decorators, and wrappers for LangChain, CrewAI, and standard Python function tools.

Does this skill log sensitive information or API keys?

No, it includes strict security guidelines and sanitization patterns to ensure that PII, passwords, and credentials are never recorded in your traces.

Tool Call Tracking

Name: Tool Call Tracking
Author: nexus-labs-automation

bynexus-labs-automation

0•

分析与监控

Instruments AI agent tool executions to provide deep visibility into function calls, performance metrics, and error handling.

Tool Call Tracking is a specialized observability skill for Claude Code that enables developers to monitor and debug how AI agents interact with external systems and environments. By implementing standardized tracing for tool executions, this skill captures critical data points like success rates, latency, and categorized errors while strictly avoiding the logging of sensitive PII or credentials. It provides pre-built patterns for sanitizing data and integrates seamlessly with popular AI frameworks like LangChain and CrewAI, ensuring that every autonomous action taken by an agent is transparent, measurable, and secure.

主要功能

01Framework-specific patterns for LangChain, CrewAI, and Python decorators

02Security-first data sanitization to prevent PII and credential leaks

03Automatic error categorization for faster debugging of agent failures

040 GitHub stars

05Standardized instrumentation for tool execution spans and latency

06Domain-specific tracking for APIs, databases, searches, and file systems

使用场景

01Identifying and resolving bottlenecks in complex multi-agent workflows

02Securing production AI agents by auditing tool inputs and outputs

03Improving agent reliability through detailed failure analysis and retry logic

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add nexus-labs-automation/agent-observability tool-call-tracking

For use in Claude.ai and ChatGPT

Download Skill