Skip to main content
The Traces tab in Contexa provides complete visibility into how your MCP servers and tools are being used. Whether youโ€™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. Youโ€™ll find two main sections:

๐ŸŸข Platform Logs

Shows all individual tool calls made across your deployed servers.

๐Ÿ“˜ 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

๐Ÿ“œ 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 โ€œView Detailโ€ on any entry opens the full trace. Screenshot2025 08 07at2 18 08PM Pn

Tool Call Detail View

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

๐Ÿ”ง Input Schema

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

๐Ÿ“ค Output Schema

Displays the toolโ€™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

  • โœ… Debug tool calls in real-time
  • ๐Ÿ”Ž Inspect malformed inputs or failures
  • ๐Ÿ“Š Analyze success rates and usage patterns
  • ๐Ÿงต Trace tool chains within a session
The Tracing section is your mission control for observabilityโ€”whether youโ€™re shipping APIs, orchestrating custom tools, or managing high-traffic MCP servers.