The Traces tab in Contexa provides complete visibility into how your MCP servers and tools are being used. Whether you’re debugging a failing tool, measuring usage trends, or just validating tool call outputs - Traces is the place to go. All tool calls are automatically logged with rich metadata, making it easy to inspect input/output schemas, status, timestamps, and session history.

​
Overview

Click on the Traces tab from the sidebar to open the tracing dashboard. You’ll find two main sections:

​
🟢 Platform Logs

Shows all individual tool calls made across your deployed servers.

​
📘 Server Session Logs

Groups calls by chat sessions, allowing you to trace the full flow of tool usage within a single conversation (ideal for multi-step agents or workflows). Screenshot2025 08 07at2 17 25PM Pn

​
📜 Tool Calls History

  • A searchable, filterable list of all tool calls
  • View includes:
    • Tool name (e.g., get-library-docs)
    • Status (Success / Failure)
    • Timestamp
    • Source MCP server
Clicking “View Detail” on any entry opens the full trace. Screenshot2025 08 07at2 18 08PM Pn

​
Tool Call Detail View

When you inspect a tool call, you’ll see:
  • Tool Name
  • Call Status
  • Timestamp
  • MCP Server Source
  • Session ID
Along with:

​
🔧 Input Schema

Shows the exact arguments sent to the tool, in JSON format.

​
📤 Output Schema

Displays the tool’s response/output, if any. This lets you validate that tools are being called correctly and returning expected results.

​
Server Session Logs

The second tab, Server Session Logs, groups tool call traces by session ID. This is especially useful when:
  • Debugging multi-step tool interactions by an agent
  • Understanding the chronological flow of calls in a single chat
  • Analyzing patterns in real-world usage

​
Use Cases

  • ✅ Debug tool calls in real-time
  • 🔎 Inspect malformed inputs or failures
  • 📊 Analyze success rates and usage patterns
  • 🧵 Trace tool chains within a session
The Tracing section is your mission control for observability—whether you’re shipping APIs, orchestrating custom tools, or managing high-traffic MCP servers.