AI Agent Documentation Sources

RevoGrid Pro publishes several documentation formats for AI agents. They all describe the same portal documentation, but serve different retrieval workflows. Start with the smallest source that answers the task, then retrieve more context only when needed.

Implementation authority

These resources provide guidance and examples. When an agent can inspect the project, local source code and public TypeScript types remain the final authority for implementation details.

Choose a source

Source	Best for	What the agent receives
llms.txt	First lookup and URL discovery	A compact Markdown index of high-signal guides, API references, demos, and links to larger resources.
llms-full.txt	Offline context or a one-shot research task	The complete normalized Markdown documentation corpus.
llms-skill.txt	Skill-aware agents that fetch only relevant references	A SKILL.md index that routes the agent to topic-specific reference files.
llms-skill/	Selective, URL-based retrieval	The same split skill index and individual Markdown references under llms-skill/references/.
revogrid-pro.skill	Installing reusable RevoGrid documentation context in a supported agent	A downloadable skill archive containing SKILL.md and its split reference files.
RevoGrid Pro MCP	Authenticated, task-oriented retrieval for Pro work	Pro-aware docs, examples, feature resolution, migration notes, and typed plugin context.

Start with llms.txt

Fetch https://pro.rv-grid.com/llms.txt when the agent needs to find the right RevoGrid page before reading it. The index favors high-signal pages, groups the remaining documentation by topic, and points to the full corpus and skill formats.

This is the best default for an agent that can follow links. It avoids placing the entire documentation corpus into context before the agent knows whether the task concerns Pivot, Gantt, filtering, or a core API.

Read https://pro.rv-grid.com/llms.txt. Find the relevant RevoGrid Pro documentation,
then fetch only the pages needed for this task before proposing an implementation.

Use the full corpus for broad context

https://pro.rv-grid.com/llms-full.txt contains every portal documentation page as normalized Markdown. Each section includes its portal URL and source path, so an agent can trace guidance back to its documentation page.

Use it for a research or offline workflow that cannot follow many links. Avoid loading it by default if the agent has a limited context window; llms.txt or the split skill is usually more efficient.

Use the split skill for selective retrieval

The split skill is designed for agents that understand SKILL.md instructions or can fetch Markdown files by URL.

• llms-skill.txt returns the root SKILL.md as plain text.
• llms-skill/ returns the same index and exposes the individual references it links to.
• revogrid-pro.skill packages that index and all reference files as a downloadable skill archive.

The index groups references by product area. Read the index first, then fetch only the referenced Markdown files that match the task. This gives an agent structured guidance without sending unrelated documentation into its context.

Read https://pro.rv-grid.com/llms-skill.txt first. Follow only the references for
the feature I am implementing, and validate API details against local TypeScript types.

Use the installable skill when supported

If your agent supports skill archives, install revogrid-pro.skill to make the split documentation available as a reusable local skill. Otherwise, use the public llms-skill.txt URL and reference links directly.

Use MCP for Pro-aware retrieval

The public files are documentation exports: they are ideal for direct HTTP retrieval and do not require a Pro token. MCP is different—it is an authenticated tool connection for agents that support MCP and need Pro-only examples, feature resolution, migration notes, or typed plugin context.

For setup, token requirements, supported clients, and example prompts, see How to use RevoGrid Pro MCP.

Recommended agent workflow

1. Fetch llms.txt or read the installed split skill index.
2. Retrieve the smallest set of relevant guides and API references.
3. When configured, query MCP for Pro-specific examples, migrations, and typed context.
4. Inspect local source and public TypeScript types before writing or changing code.
5. Keep the implementation aligned with the retrieved documentation and the local project version.

This sequence gives agents quick discovery, focused context, and a final implementation check against the code they are changing.