Scout Trails
See what your AI coding agents are actually doing.
AI agents are becoming part of the engineering team. Scout Trails gives you visibility into how they work. See every session, tool call, and dollar spent across your team, all in one place.
Works with Claude Code, and other OpenTelemetry-compatible agents.
Why it matters
Your agents are working. Now you can see the work.
Autonomous agents are black boxes. They make changes, call tools, and touch your codebase, and when something goes wrong, there's no record of what they actually did.
Scout Trails gives engineering teams a shared view of agent activity across developers, repositories, and sessions.
Understand adoption
See how AI agents are being used across your engineering team.
Investigate sessions
Drill into individual sessions and see the tools and actions behind them.
Track usage and cost
Understand where tokens and dollars are going as agent adoption grows.
Spot patterns
See which workflows, tools, and agents are actually becoming part of development.
Audit what happened
See exactly what an agent did when a routine behaves unexpectedly, without reconstructing it after the fact.
Individual visibility doesn't scale to team visibility.
As AI coding agents become part of everyday development, engineering leaders need more than individual terminal histories.
Scout brings agent activity into one place so you can see how agents are being used across your team, without asking every developer to reconstruct what their agent did.
And because token spend adds up fast, Scout shows you where it's going before a surprise shows up on the invoice.
Every session. Every tool call. Every dollar. One view.
Get Started
Five environment variables. That's it.
No downloaded binary. No wrapper process.
Point your existing AI coding agent at Scout's hosted ingest service by configuration alone, and start sending agent telemetry over OpenTelemetry.
The bigger picture
Your AI knows the code. Scout knows production.
AI coding agents are getting better at understanding your codebase. But code alone can't tell them what actually happened in production.
Scout already can.
Production traces show what a request actually did. Performance history shows whether behavior is new, recurring, or getting worse. Scout's MCP integration makes that context available inside the AI tools developers already use.
Scout Trails extends that visibility to the other side of the development loop.
One system. Both sides of AI development.
Understand the agent
See the sessions, tools, usage, and cost behind AI-assisted development.
Understand production
Give AI access to real application performance, traces, and historical context.
Close the loop
Move from "what did the agent do?" to "what did the code do in production?" without asking AI to guess.
FAQ
Agent Monitoring FAQ
What does Scout's agent monitoring track?
It gives your team visibility into what AI coding agents are actually doing: every session, every tool call, and the cost behind it, in one shared view instead of scattered terminal histories that disappear the moment a developer closes their editor.
Which AI coding agents does it work with?
Claude Code today, plus any other OpenTelemetry-compatible coding agent. Setup is configuration only: point your agent's telemetry at Scout's hosted ingest endpoint with five environment variables, no downloaded binary or wrapper process required.
When is it available?
It's available now in beta. Sign up at scoutmonitoring.io to get started. We're still shaping the product with early users, so feedback is very welcome.
Is it free during beta?
Yes, beta access is free while we validate the product with early users. We'll share pricing details well before any beta features move to general availability.
How is this different from Scout's MCP server?
They're two sides of the same loop. Scout's MCP server gives your AI coding agent access to Scout's production data, traces, errors, and performance history, while it's working. Agent monitoring is the reverse: it gives you visibility into what the agent itself did, so you have both what happened in your code and what happened in production, in one place.
How is this different from Claude Code's built-in history or usage reporting?
Claude Code's history lives locally, tied to one developer's machine and session. Scout Trails pulls session activity, tool calls, and cost into one shared view across your whole engineering team, so you're not asking every developer to dig through their own local history to piece together what an agent did.
How does Scout Trails work with Scout APM?
They cover both sides of the development loop. Scout APM already shows you what happened in production: traces, errors, performance history. Scout Trails shows you what your AI agents did in the codebase to get there. Used together, you can go from "the agent changed this" to "here's what that did in production" without stitching the two together yourself.
Do I need Scout APM to use Trails?
No. Scout Trails is a separate product with its own signup, no Scout APM account required. If you do use Scout APM, the two are built to work together, but Trails works fine on its own.