> ## Documentation Index
> Fetch the complete documentation index at: https://docs.withgauge.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Developer documentation

> Measure how coding agents experience, choose, and use developer tools.

Gauge runs coding agents in real repositories so you can see what they choose, where they struggle, and whether a change improves the outcome.

Use the full [Gauge web product](https://agents.withgauge.com) or the Gauge CLI to onboard, configure measurements and run presets, launch runs, and inspect results. Both interfaces provide the same core Coding Agent Optimization workflows.

Use Gauge to measure two parts of agent-led growth:

* **[Agent Preference](/agents/concepts/agent-preference)** shows how often agents recommend or adopt your product for a realistic developer task.
* **[Agent Experience](/agents/concepts/agent-experience)** checks whether agents can complete a task with your product and why they pass or fail.

<CardGroup cols={3}>
  <Card title="How Gauge works" icon="diagram-project" href="/agents/concepts/how-gauge-works">
    Follow a measurement from its prompt and run preset through execution, judging, and analysis.
  </Card>

  <Card title="CLI guide" icon="terminal" href="/agents/guides/cli">
    Configure measurements, launch runs, and query results from your terminal or CI.
  </Card>

  <Card title="API reference" icon="brackets-curly" href="/agents/api-reference/overview">
    Integrate with the bearer-token API that powers the Gauge CLI.
  </Card>
</CardGroup>

## Core concepts

| Concept     | What it controls                                                                        |
| ----------- | --------------------------------------------------------------------------------------- |
| Measurement | The question or task Gauge asks an agent to complete and the criteria used to judge it. |
| Run preset  | The repository, agent and model, persona, skills, and MCP servers used for a run.       |
| Cycle       | How often attached measurements and run presets expand into new runs.                   |
| Run         | One coding-agent session in an isolated workspace.                                      |
| Experiment  | A set of candidate changes tested against the same task and context.                    |