Scout Trails Docs

Overview

Scout Trails records what your LLM coding agents do. Agents such as Claude Code and Codex CLI send their built-in OpenTelemetry data to the Scout Trails ingest endpoint, and the dashboard turns it into sessions, tool calls, cost and trends for your organization. There is nothing to install: you point each agent at the endpoint with a few lines of configuration and an ingest key. This section explains the core concepts and walks through each page of the dashboard.

Core concepts

Organization

An organization is the top-level container for everything in Scout Trails: its members, teams, actors, keys and all the telemetry they send. Data never crosses organizations. Every member of an organization can read all of that organization’s data; roles only control who can change settings. A person can belong to more than one organization and switch between them from the organization menu in the dashboard header.

Member

A member is a person who can sign in to the dashboard. A member has an email address, a password and one role in each organization they belong to:

  • member — reads all of the organization’s data, invites people, and manages actors, ingest keys and API keys.
  • admin — everything a member can do, plus renaming the organization, managing teams, and changing members’ roles or removing members.

People join an organization by accepting an invitation, and always join as member. See Members.

Actor

An actor is anything that sends telemetry to Scout Trails: your laptop, a teammate’s workstation, a CI runner, an autonomous agent. Every actor has a display name, a kind (Human or Agent), an optional role label, and at most one ingest key. Each tool call and model call is attributed to the actor whose key sent it, and that actor’s display name is what the dashboard shows in its Actor columns.

An actor is not the same as a member. An actor can be tied to a member’s account, but it doesn’t have to be: a CI runner is an actor with no login. A member who signs in to the dashboard sends nothing until they create an actor and configure an agent with its key.

Give each machine or agent its own actor. The Actor column and the actor options in the Scope filter are only as precise as that split. See Actors.

Team

A team is a named group of actors within an organization. Teams don’t control access, since every member already sees all of the organization’s data. They exist so you can narrow any dashboard page to a group of actors with the Scope filter. See Teams.

Session

A session is one run of a coding agent, identified by the session ID the agent reports. Scout Trails builds sessions from tool calls, so a session appears once at least one of its tool calls arrives. When an agent resumes an earlier conversation it reuses the same session ID, so one session can span days or weeks. See Sessions.

Tool call

A tool call is one invocation of a tool by an agent, such as reading a file, running a shell command, fetching a URL or calling an MCP server. Each tool call has a name, a type, a category, a status (ok, error or denied), a duration when the tool ran, and the arguments and result the agent reported. See Tool calls.

Model call and cost

A model call is one request the agent made to its language model, with token counts and a dollar cost. Cost figures in the dashboard come from model calls. Claude Code reports model calls with their cost. Codex CLI reports tool calls only, so Codex sessions show no cost.

Ingest key

An ingest key is the credential an agent sends with its telemetry, as a Bearer token in the Authorization header. It identifies the actor, and through the actor the organization, that the data belongs to. Each actor holds at most one key. A new key works as soon as it is created, and a revoked key stops working immediately. Ingest keys only send data; they cannot read anything. See Ingest keys.

API key

An API key is a secret credential for reading your organization’s data programmatically, through the REST API or an MCP client. API keys belong to the organization. An ingest key does not work in their place. See API keys.

How the pieces fit

  1. You create an actor in the dashboard and copy its ingest key.
  2. You configure an agent to send OpenTelemetry data to the ingest endpoint, https://ingest.scoutmonitoring.io:443, with that key.
  3. Scout Trails attributes each tool call and model call to the actor and its organization.
  4. Every member of the organization sees the data in the dashboard at https://ui.scoutmonitoring.io, and can read it through the API or an MCP client with an API key.

Getting started

  1. Sign in to the dashboard, or accept the invitation someone sent you. See Signing in and your account.
  2. Create an actor for the machine you want to instrument and copy its ingest key. See Actors.
  3. Configure your agent with the ingest key. See Agent setup.
  4. Run your agent for a minute, then open Sessions in the dashboard to see the session. See Sessions.
  5. Optionally, create an API key to query your data from scripts or from an MCP-capable assistant. See API and MCP.

In this section

  • Dashboard basics — navigation, the time window and Scope filters, and the Overview page.
  • Sessions — the sessions list and the session detail page.
  • Tool calls — the tool calls list and the tool call detail page.
  • Trends — activity, cost, outlier sessions and top tools over a window.
  • Signing in and your account — signing in, resetting your password, accepting an invitation and personal settings.