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

# llms.txt and llms-full.txt

> Understand CloudEval's public machine-readable context files and when to use each one.

CloudEval publishes two public machine-readable context files on the main product domain. They are part of the supported developer and agent experience.

<Info>These files are public and supported. They are meant to help agents and developer tools route themselves quickly, not replace capability checks or the main docs.</Info>

## Available files

<Columns cols={2}>
  <Card title="llms.txt" icon="file-text" href="https://cloudeval.ai/llms.txt">
    Start here when you need a short product summary, core links, and the high-level CLI surface.
  </Card>

  <Card title="llms-full.txt" icon="files" href="https://cloudeval.ai/llms-full.txt">
    Use this when you need detailed workflows, command guidance, trust boundaries, and limitations.
  </Card>
</Columns>

<Note>Use the canonical `llms.txt` and `llms-full.txt` URLs in automation.</Note>

## What these files cover

The two files are designed to help developers and agents understand CloudEval without scraping the UI.

They cover:

* what CloudEval does
* current provider and IaC support
* current public pricing and disabled commercial tiers
* diagram and workspace review concepts
* the CLI command surface
* terminal UI versus automation guidance
* profile-aware setup and config
* diagnostics such as `status` and `doctor`
* searchable local session history and session resume workflows
* MCP client setup patterns, server usage, focused toolsets, resources, and prompts
* headless diagram image download commands and auth boundaries
* common workflows
* real import examples for local ARM JSON and GitHub template URLs
* GitHub URL project creation and redirect behavior
* visual review concepts for architecture maps, dependency views, Cloud Evaluation reports, issue lists, CLI handoff, and MCP setup
* report context patterns for Brief, Action Plan, Architecture & Dependencies, Security & Compliance, Cost & FinOps, Resources, and Evidence & Progress
* report export and handoff guidance for PDF, Markdown, JSON, scoped summaries, evidence packs, and external AI tools
* source snapshot, model context, and planned custom LLM gateway boundaries
* agent usage rules
* data grounding expectations
* limitations and freshness guidance
* public URLs such as the sitemap and search index

For launch, the public search index should include docs, release notes, AEO pages, and machine-readable context files. It should not index sample blog content, redirect-only pages, or generic tutorials that are not intended as launch entrypoints.

## How they are discovered

CloudEval publishes these files as public URLs on the product site:

* `https://cloudeval.ai/llms.txt`
* `https://cloudeval.ai/llms-full.txt`
* `https://cloudeval.ai/search-index.json`
* `https://cloudeval.ai/sitemap.xml`
* `https://cloudeval.ai/robots.txt`

The frontend also includes `llms.txt` and `llms-full.txt` in the sitemap, and the robots file points agents at both endpoints directly.

## When to use each file

Start with `llms.txt` when you want:

* a short product summary
* the main public URLs
* the current high-level CLI surface
* a quick pointer to the detailed file

Fetch `llms-full.txt` when you want:

* full CLI examples
* terminal UI guidance
* profile, diagnostics, sessions, and MCP details
* MCP setup patterns for Codex, Cursor, Claude Code, VS Code, generic JSON-configured clients, and Ollama-powered agent hosts
* auth and automation guidance
* ASK versus AGENT usage expectations
* grounding and trust model details
* current limitations and support boundaries

## What these files do not replace

These files are helpful context, but they are not the only source of truth.

* Use `cloudeval capabilities --format json` before relying on a command in automation.
* Use [Create a project from a GitHub URL](/quickstart/create-a-project-from-a-github-url) for the browser-first URL import flow.
* Use [Reports](/concepts/reports) when interpreting Cloud Evaluation report tabs, evidence, audience modes, and exports.
* Use [Report types and statuses](/reference/report-types-and-statuses) when distinguishing report payloads from report tabs.
* Use [MCP client setup](/reference/mcp-client-setup) when connecting Codex, Cursor, Claude Code, VS Code, or another MCP-compatible client.
* Use [Headless diagram image downloads](/reference/headless-diagram-image-downloads) when an agent needs architecture or dependency image bytes directly.
* Use [Data and model boundaries](/reference/data-and-model-boundaries) when deciding whether the current hosted model path is acceptable for private source or cloud evidence.
* Prefer current app behavior over an older copied example.
* Do not assume provider parity just because a provider name appears in the UI or a file.
* Do not assume custom LLM gateway routing or bring-your-own model key support is available in the self-serve Pro product today.
* Treat the `Last updated` field as a hint, not a guarantee that every example matches the current release.

## Public boundaries

The `llms` files are public documentation. They should describe what exists today, call out limitations honestly, and avoid exposing private operational detail.

That means:

* no speculative product claims
* no hidden APIs presented as public surface area
* no unsupported provider promises
* no stale public pricing
* no sample or tutorial content in the public search index unless the page is intentionally launch-ready

## Next step

Use [Agent and automation rules](/reference/agent-and-automation-rules) if you are consuming these files from tools or agents, [MCP client setup](/reference/mcp-client-setup) for client-specific setup, or [CLI overview](/reference/cli-overview) if you want the terminal model first.